
!pip install keras_tuner
!pip install orbit-ml
!pip install bayesian-optimization
!pip install pmdarima
!pip install prophet

Collecting keras_tuner
  Downloading keras_tuner-1.4.7-py3-none-any.whl (129 kB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 129.1/129.1 kB 1.6 MB/s eta 0:00:00
Requirement already satisfied: keras in /usr/local/lib/python3.10/dist-packages (from keras_tuner) (2.15.0)
Requirement already satisfied: packaging in /usr/local/lib/python3.10/dist-packages (from keras_tuner) (24.0)
Requirement already satisfied: requests in /usr/local/lib/python3.10/dist-packages (from keras_tuner) (2.31.0)
Collecting kt-legacy (from keras_tuner)
  Downloading kt_legacy-1.0.5-py3-none-any.whl (9.6 kB)
Requirement already satisfied: charset-normalizer<4,>=2 in /usr/local/lib/python3.10/dist-packages (from requests->keras_tuner) (3.3.2)
Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.10/dist-packages (from requests->keras_tuner) (3.7)
Requirement already satisfied: urllib3<3,>=1.21.1 in /usr/local/lib/python3.10/dist-packages (from requests->keras_tuner) (2.0.7)
Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.10/dist-packages (from requests->keras_tuner) (2024.2.2)
Installing collected packages: kt-legacy, keras_tuner
Successfully installed keras_tuner-1.4.7 kt-legacy-1.0.5
Collecting pmdarima
  Downloading pmdarima-2.0.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.manylinux_2_28_x86_64.whl (2.1 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 2.1/2.1 MB 5.4 MB/s eta 0:00:00
Requirement already satisfied: joblib>=0.11 in /usr/local/lib/python3.10/dist-packages (from pmdarima) (1.4.2)
Requirement already satisfied: Cython!=0.29.18,!=0.29.31,>=0.29 in /usr/local/lib/python3.10/dist-packages (from pmdarima) (3.0.10)
Requirement already satisfied: numpy>=1.21.2 in /usr/local/lib/python3.10/dist-packages (from pmdarima) (1.25.2)
Requirement already satisfied: pandas>=0.19 in /usr/local/lib/python3.10/dist-packages (from pmdarima) (2.0.3)
Requirement already satisfied: scikit-learn>=0.22 in /usr/local/lib/python3.10/dist-packages (from pmdarima) (1.2.2)
Requirement already satisfied: scipy>=1.3.2 in /usr/local/lib/python3.10/dist-packages (from pmdarima) (1.11.4)
Requirement already satisfied: statsmodels>=0.13.2 in /usr/local/lib/python3.10/dist-packages (from pmdarima) (0.14.2)
Requirement already satisfied: urllib3 in /usr/local/lib/python3.10/dist-packages (from pmdarima) (2.0.7)
Requirement already satisfied: setuptools!=50.0.0,>=38.6.0 in /usr/local/lib/python3.10/dist-packages (from pmdarima) (67.7.2)
Requirement already satisfied: packaging>=17.1 in /usr/local/lib/python3.10/dist-packages (from pmdarima) (24.0)
Requirement already satisfied: python-dateutil>=2.8.2 in /usr/local/lib/python3.10/dist-packages (from pandas>=0.19->pmdarima) (2.8.2)
Requirement already satisfied: pytz>=2020.1 in /usr/local/lib/python3.10/dist-packages (from pandas>=0.19->pmdarima) (2023.4)
Requirement already satisfied: tzdata>=2022.1 in /usr/local/lib/python3.10/dist-packages (from pandas>=0.19->pmdarima) (2024.1)
Requirement already satisfied: threadpoolctl>=2.0.0 in /usr/local/lib/python3.10/dist-packages (from scikit-learn>=0.22->pmdarima) (3.5.0)
Requirement already satisfied: patsy>=0.5.6 in /usr/local/lib/python3.10/dist-packages (from statsmodels>=0.13.2->pmdarima) (0.5.6)
Requirement already satisfied: six in /usr/local/lib/python3.10/dist-packages (from patsy>=0.5.6->statsmodels>=0.13.2->pmdarima) (1.16.0)
Installing collected packages: pmdarima
Successfully installed pmdarima-2.0.4


import pandas as pd
import numpy as np
from datetime import datetime
import itertools
import matplotlib.pyplot as plt
import seaborn as sns
from tqdm import tqdm
import datetime
import scipy.stats
from scipy.stats import norm
import matplotlib.dates as mdates
import os

from math import sqrt
import scipy.stats as stats
from scipy.stats import randint, uniform

import statsmodels
import statsmodels.api as sm
from statsmodels.tsa.seasonal import seasonal_decompose
from statsmodels.tsa.stattools import acf, pacf
from statsmodels.graphics.tsaplots import plot_acf, plot_pacf
from statsmodels.tsa.stattools import adfuller
from statsmodels.tsa.statespace.sarimax import SARIMAX
from statsmodels.stats.diagnostic import acorr_ljungbox
from statsmodels.tsa.arima.model import ARIMA
from statsmodels.tsa.stattools import kpss

from sklearn.model_selection import train_test_split
from sklearn.metrics import mean_squared_error, mean_absolute_error
from sklearn.model_selection import RandomizedSearchCV
from sklearn.preprocessing import MinMaxScaler
from sklearn.base import BaseEstimator

import torch
from tensorflow import keras
import kerastuner as kt
from keras.callbacks import EarlyStopping, ModelCheckpoint
#from bayes_opt import BayesianOptimization


from pmdarima import auto_arima

#from orbit.models import DLT
#from orbit.diagnostics.plot import plot_predicted_data
#from orbit.utils.params_tuning import grid_search_orbit
#from orbit.utils.features import make_fourier_series_df
#from orbit.utils.params_tuning import generate_param_args_list

# from prophet import Prophet

from google.colab import drive
drive.mount('/content/drive')

Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount("/content/drive", force_remount=True).


df_2020_Q1 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/Divvy_Trips_2020_Q1.csv")

df_2020_Q1.head()


from google.colab import drive
drive.mount('/content/drive')

Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount("/content/drive", force_remount=True).


df_2020_04 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202004-divvy-tripdata.csv")
df_2020_04.head()


df_2020_05 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202005-divvy-tripdata.csv")
df_2020_05.head()


df_2020_06 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202006-divvy-tripdata.csv")
df_2020_06.head()


df_2020_07 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202007-divvy-tripdata.csv")
df_2020_07.head()


df_2020_08 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202008-divvy-tripdata.csv")
df_2020_08.head()


df_2020_09 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202009-divvy-tripdata.csv")
df_2020_09.head()


df_2020_10 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202010-divvy-tripdata.csv")
df_2020_10.head()


df_2020_11 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202011-divvy-tripdata.csv")
df_2020_11.head()


df_2020_12 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202012-divvy-tripdata.csv")
df_2020_12.head()


df_2021_01 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202101-divvy-tripdata.csv")
df_2021_01.head()


df_2021_02 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202102-divvy-tripdata.csv")
df_2021_02.head()


df_2021_03 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202103-divvy-tripdata.csv")
df_2021_03.head()


df_2021_04 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202104-divvy-tripdata.csv")
df_2021_04.head()


df_2021_05 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202105-divvy-tripdata.csv")
df_2021_05.head()


df_2021_06 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202106-divvy-tripdata.csv")
df_2021_06.head()


df_2021_07 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202107-divvy-tripdata.csv")
df_2021_07.head()


df_2021_08 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202108-divvy-tripdata.csv")
df_2021_08.head()


df_2021_09 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202109-divvy-tripdata.csv")
df_2021_09.head()


df_2021_10 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202110-divvy-tripdata.csv")
df_2021_10.head()


df_2021_11 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202111-divvy-tripdata.csv")
df_2021_11.head()


df_2021_12 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202112-divvy-tripdata.csv")
df_2021_12.head()


df_2022_01 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202201-divvy-tripdata.csv")
df_2022_01.head()


df_2022_02 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202202-divvy-tripdata.csv")
df_2022_02.head()


df_2022_03 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202203-divvy-tripdata.csv")
df_2022_03.head()


df_2022_04 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202204-divvy-tripdata.csv")
df_2022_04.head()


df_2022_05 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202205-divvy-tripdata.csv")
df_2022_05.head()


df_2022_06 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202206-divvy-tripdata.csv")
df_2022_06.head()


df_2022_07 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202207-divvy-tripdata.csv")
df_2022_07.head()


df_2022_08 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202208-divvy-tripdata.csv")
df_2022_08.head()


df_2022_09 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202209-divvy-publictripdata.csv")
df_2022_09.head()


df_2022_10 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202210-divvy-tripdata.csv")
df_2022_10.head()


df_2022_11 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202211-divvy-tripdata.csv")
df_2022_11.head()


df_2022_12 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202212-divvy-tripdata.csv")
df_2022_12.head()


df_2023_01 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202301-divvy-tripdata.csv")
df_2023_01.head()


df_2023_02 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202302-divvy-tripdata.csv")
df_2023_02.head()


df_2023_03 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202303-divvy-tripdata.csv")
df_2023_03.head()


df_2023_04 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202304-divvy-tripdata.csv")
df_2023_04.head()


df_2023_05 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202305-divvy-tripdata.csv")
df_2023_05.head()


df_2023_06 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202306-divvy-tripdata.csv")
df_2023_06.head()


df_2023_07 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202307-divvy-tripdata.csv")
df_2023_07.head()


df_2023_08 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202308-divvy-tripdata.csv")
df_2023_08.head()


df_2023_09 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202309-divvy-tripdata.csv")
df_2023_09.head()


df_2023_10 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202310-divvy-tripdata.csv")
df_2023_10.head()


df_2023_11 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202311-divvy-tripdata.csv")
df_2023_11.head()


df_2023_12 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202312-divvy-tripdata.csv")
df_2023_12.head()


df_2024_01 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202401-divvy-tripdata.csv")
df_2024_01.head()


df_2024_02 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202402-divvy-tripdata.csv")
df_2024_02.head()


df_2024_03 = pd.read_csv("/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/202403-divvy-tripdata.csv")
df_2024_03.head()


df_bike = pd.concat([df_2020_Q1, df_2020_04, df_2020_05,
                         df_2020_06, df_2020_07, df_2020_08,
                         df_2020_09, df_2020_10, df_2020_11,
                         df_2020_12, df_2021_01, df_2021_02,
                         df_2021_03, df_2021_04, df_2021_05,
                         df_2021_06, df_2021_07, df_2021_08,
                         df_2021_09, df_2021_10, df_2021_11,
                         df_2021_12, df_2022_01, df_2022_02,
                         df_2022_03, df_2022_04, df_2022_05,
                         df_2022_06, df_2022_07, df_2022_08,
                         df_2022_09, df_2022_10, df_2022_11,
                         df_2022_12, df_2023_01, df_2023_02,
                         df_2023_03, df_2023_04, df_2023_05,
                         df_2023_06, df_2023_07, df_2023_08,
                         df_2023_09, df_2023_10, df_2023_11,
                         df_2023_12, df_2024_01, df_2024_02,
                         df_2024_03], ignore_index=True)


# Assuming 'df_bike' is the DataFrame you want to save
df_bike.to_csv('/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/bike_data.csv', index=False)


df_bike = pd.read_csv('/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/bike_data_cleaned.csv')
df_bike.head()


df_bike['started_at'] = pd.to_datetime(df_bike['started_at'])
df_bike['ended_at'] = pd.to_datetime(df_bike['ended_at'])


df_bike = df_bike[(df_bike['started_at'] >= '2020-03-01') & (df_bike['started_at'] <= '2024-03-31')]


class_df = df_bike.loc[df_bike['rideable_type'] == 'classic_bike']
elect_df = df_bike.loc[df_bike['rideable_type'] == 'electric_bike']
dock_df = df_bike.loc[df_bike['rideable_type'] == 'docked_bike']


class_df.isnull().sum()


elect_df.isnull().sum()


dock_df.isnull().sum()


class_df.head()


## Classic Bike
class_bike_demand = class_df.resample('D', on='started_at')['ride_id'].count()
class_bike_demand = class_bike_demand.to_frame(name='classic_bike_rentals')

# Electric Bike
elect_bike_demand = elect_df.resample('D', on='started_at')['ride_id'].count()
elect_bike_demand = elect_bike_demand.to_frame(name='elect_bike_rentals')

# Docked Bike
dock_bike_demand = dock_df.resample('D', on='started_at')['ride_id'].count()
dock_bike_demand = dock_bike_demand.to_frame(name='dock_bike_rentals')

# class_bike_demand.to_csv('/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/classic_bike_data.csv', index=True)
# elect_bike_demand.to_csv('/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/electric_bike_data.csv', index=True)
# dock_bike_demand.to_csv('/content/drive/MyDrive/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/docked_bike_data.csv', index=True)


class_df = pd.read_csv('//content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/classic_bike_data.csv')
elect_df = pd.read_csv('/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Divvy CSV/electric_bike_data.csv')


class_df['started_at']=pd.to_datetime(class_df['started_at'])
class_df['day']=class_df['started_at'].dt.day_name()

elect_df['started_at']=pd.to_datetime(elect_df['started_at'])
elect_df['day']=elect_df['started_at'].dt.day_name()


class_df = class_df.set_index('started_at')
elect_df = elect_df.set_index('started_at')


average_rentals_classic = class_df.groupby('day')['classic_bike_rentals'].mean().reindex([
    'Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday', 'Sunday'
])
sns.barplot(x=average_rentals_classic.index, y=average_rentals_classic.values, color='skyblue')
plt.xlabel('Day of the Week')
plt.ylabel('Average Bikes Rented')
plt.title('Average Classic Bike Rentals by Day of the Week')
plt.xticks(rotation=45)
plt.show()


average_rentals_e = elect_df.groupby('day')['elect_bike_rentals'].mean().reindex([
    'Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday', 'Sunday'
])
sns.barplot(x=average_rentals_e.index, y=average_rentals_e.values, color='skyblue')
plt.xlabel('Day of the Week')
plt.ylabel('Average Bikes Rented')
plt.title('Average Electric Bike Rentals by Day of the Week')
plt.xticks(rotation=45)  # Rotate labels to improve readability
plt.show()


# Displaying basic plots to visualize any obvious patterns and trends
class_df['classic_bike_rentals'].plot(kind='box')
plt.title('Box Plot of Daily Classicc Bike Rentals')
plt.grid(True)
plt.show()


plt.figure(figsize=(10, 6))
plt.plot(class_df.index, class_df['classic_bike_rentals'], color = 'red')

plt.xlabel('Date', fontsize=14)
plt.ylabel('Classic Bike Rentals', fontsize=14)
plt.title('Daily Demand of Classic Bike Rentals', fontsize=16)

plt.xticks(rotation=45)
plt.show()


# Decomposing the time series to observe trend, seasonality, and residuals
decomposition = seasonal_decompose(class_df['classic_bike_rentals'], model='additive', period=365)

# Plotting the decomposed components
plt.figure(figsize=(14, 10))
plt.subplot(411)
plt.title('Decomposing the Daily Demand of Classic Bike Rentals to Trend, Seasonality, and Residuals', fontsize=16)
plt.plot(decomposition.observed, label='Original')
plt.legend(loc='best')
plt.subplot(412)
plt.plot(decomposition.trend, label='Trend')
plt.legend(loc='best')
plt.subplot(413)
plt.plot(decomposition.seasonal, label='Seasonal')
plt.legend(loc='best')
plt.subplot(414)
plt.plot(decomposition.resid, label='Residual')
plt.legend(loc='best')
plt.tight_layout()
plt.show()


# Displaying basic plots to visualize any obvious patterns and trends
elect_df['elect_bike_rentals'].plot(kind='box')
plt.title('Box Plot of Daily Electric Bike Rentals')
plt.grid(True)
plt.show()


plt.figure(figsize=(10, 6))
plt.plot(elect_df.index, elect_df['elect_bike_rentals'], color = 'blue')

plt.xlabel('Date', fontsize=14)
plt.ylabel('Electric Bike Rentals', fontsize=14)
plt.title('Daily Demand of Electric Bike Rentals', fontsize=16)

plt.xticks(rotation=45)

plt.show()


# Decomposing the time series to observe trend, seasonality, and residuals
decomposition = seasonal_decompose(elect_df['elect_bike_rentals'], model='additive', period=365)

# Plotting the decomposed components
plt.figure(figsize=(14, 10))
plt.subplot(411)
plt.title('Decomposing the Daily Demand of Electric Bike Rentals to Trend, Seasonality, and Residuals', fontsize=16)
plt.plot(decomposition.observed, label='Original')
plt.legend(loc='best')
plt.subplot(412)
plt.plot(decomposition.trend, label='Trend')
plt.legend(loc='best')
plt.subplot(413)
plt.plot(decomposition.seasonal, label='Seasonal')
plt.legend(loc='best')
plt.subplot(414)
plt.plot(decomposition.resid, label='Residual')
plt.legend(loc='best')
plt.tight_layout()
plt.show()


class_df['log_classic_bike_rentals'] = np.log(class_df['classic_bike_rentals'])
plt.figure(figsize=(10, 6))
plt.plot(class_df.index, class_df['log_classic_bike_rentals'], color = 'red')

plt.xlabel('Date', fontsize=14)
plt.ylabel('Classic Bike Rentals', fontsize=14)
plt.title('Daily Demand of Classic Bike Rentals (Log)', fontsize=16)

plt.xticks(rotation=45)

plt.show()


## test stationary
adf_result = adfuller(class_df[class_df.columns[-1]])
adf_result

(-2.467954921769883,
 0.12346716809436559,
 22,
 1192,
 {'1%': -3.4358478507713777,
  '5%': -2.863967751840151,
  '10%': -2.568062580992973},
 905.6029861325433)


# Differencing the series
class_df['Differenced'] = class_df[class_df.columns[-1]].diff().dropna()

plt.figure(figsize=(10, 6))
plt.plot(class_df.index, class_df['Differenced'], color = 'red')

plt.xlabel('Date', fontsize=14)
plt.ylabel('Classic Bike Rentals', fontsize=14)
plt.title('Daily Demand of Classic Bike Rentals (Difference)', fontsize=16)

plt.xticks(rotation=45)
plt.show()


# Perform the Augmented Dickey-Fuller test on the differenced data
adf_result_diff = adfuller(class_df['Differenced'].dropna())
adf_result_diff

(-10.859590757061333,
 1.4578526370040222e-19,
 19,
 1194,
 {'1%': -3.4358386217286885,
  '5%': -2.8639636801920956,
  '10%': -2.5680604125597277},
 908.2749670259309)


class_df = class_df[class_df['Differenced'].notna()]


# identify potential parameters by trying ACF and PACF
fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(12, 8))

# Plot the Autocorrelation Function (ACF)
plot_acf(class_df['Differenced'].dropna(), lags=40, ax=ax1)

# Plot the Partial Autocorrelation Function (PACF)
plot_pacf(class_df['Differenced'].dropna(), lags=40, ax=ax2)
plt.show()


elect_df


elect_df['log_elect_bike_rentals'] = np.log(elect_df['elect_bike_rentals'])
plt.figure(figsize=(10, 6))
plt.plot(elect_df.index, elect_df['log_elect_bike_rentals'], color = 'blue')

plt.xlabel('Date', fontsize=14)
plt.ylabel('Electric Bike Rentals', fontsize=14)
plt.title('Daily Demand of Electric Bike Rentals (Log)', fontsize=16)

plt.xticks(rotation=45)

plt.show()

/usr/local/lib/python3.10/dist-packages/pandas/core/arraylike.py:396: RuntimeWarning: divide by zero encountered in log
  result = getattr(ufunc, method)(*inputs, **kwargs)


elect_df = elect_df[elect_df['log_elect_bike_rentals'] != float('-inf')]
elect_df


## test stationary
adf_result = adfuller(elect_df['log_elect_bike_rentals'])
adf_result

(-2.8839979459818124,
 0.047240713585156066,
 19,
 1325,
 {'1%': -3.435294916169133,
  '5%': -2.863723787379918,
  '10%': -2.5679326566037735},
 1353.1334161865007)


kpss_stats=kpss(elect_df[elect_df.columns[-1]], 'c')
kpss_stats

<ipython-input-95-32e14d2bb934>:1: InterpolationWarning: The test statistic is outside of the range of p-values available in the
look-up table. The actual p-value is smaller than the p-value returned.

  kpss_stats=kpss(elect_df[elect_df.columns[-1]], 'c')

(1.1480004388301237,
 0.01,
 20,
 {'10%': 0.347, '5%': 0.463, '2.5%': 0.574, '1%': 0.739})


# identify potential parameters by trying ACF and PACF
fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(12, 8))

# Plot the Autocorrelation Function (ACF)
plot_acf(elect_df['log_elect_bike_rentals'].dropna(), lags=40, ax=ax1)

# Plot the Partial Autocorrelation Function (PACF)
plot_pacf(elect_df['log_elect_bike_rentals'].dropna(), lags=40, ax=ax2)

plt.show()


# Differencing the series
elect_df['Differenced'] = elect_df[elect_df.columns[-1]].diff().dropna()

plt.figure(figsize=(10, 6))
plt.plot(elect_df.index, elect_df['Differenced'], color = 'blue')

plt.xlabel('Date', fontsize=14)
plt.ylabel('Electric Bike Rentals', fontsize=14)
plt.title('Daily Demand of Electric Bike Rentals (Difference)', fontsize=16)

plt.xticks(rotation=45)

plt.show()
# Perform the Augmented Dickey-Fuller test on the differenced data
adf_result_diff = adfuller(elect_df['Differenced'].dropna())

adf_result_diff

<ipython-input-97-f311b1f68697>:2: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  elect_df['Differenced'] = elect_df[elect_df.columns[-1]].diff().dropna()

(-11.562302061590916,
 3.2480762955317108e-21,
 18,
 1325,
 {'1%': -3.435294916169133,
  '5%': -2.863723787379918,
  '10%': -2.5679326566037735},
 1357.5456802509739)


# Differencing the series to achieve stationarity
elect_df['diff_elect_bike_rentals'] = elect_df['elect_bike_rentals'].diff()

# Dropping any NaN values that result from differencing
bike_data_diff = elect_df.dropna()

# Plotting the differenced data
plt.figure(figsize=(10, 6))
plt.plot(bike_data_diff.index, bike_data_diff['diff_elect_bike_rentals'], marker='o', linestyle='-')
plt.title('Differenced Electric Bike Rentals Over Time')
plt.xlabel('Date')
plt.ylabel('Differenced Bike Rentals')
plt.grid(True)
plt.show()

# Re-running the Augmented Dickey-Fuller test on the differenced data
adf_result_diff = adfuller(bike_data_diff['diff_elect_bike_rentals'])
adf_statistic_diff, adf_p_value_diff, adf_critical_values_diff = adf_result_diff[0], adf_result_diff[1], adf_result_diff[4]

adf_statistic_diff, adf_p_value_diff, adf_critical_values_diff

<ipython-input-98-6f5834308715>:2: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  elect_df['diff_elect_bike_rentals'] = elect_df['elect_bike_rentals'].diff()

(-9.471577896980115,
 4.092750264125932e-16,
 {'1%': -3.4353061595148073,
  '5%': -2.863728748588099,
  '10%': -2.567935298646666})


# identify potential parameters by trying ACF and PACF
fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(12, 8))

# Plot the Autocorrelation Function (ACF)
plot_acf(elect_df['Differenced'].dropna(), lags=40, ax=ax1)

# Plot the Partial Autocorrelation Function (PACF)
plot_pacf(elect_df['Differenced'].dropna(), lags=40, ax=ax2)
plt.show()


total_days_class = (class_df.index.max() - class_df.index.min()).days + 1
print(f"Total number of days of Classic Bike Data: {total_days_class}")

Total number of days of Classic Bike Data: 1214


training_days_class = int(total_days_class * 0.8)
cutoff_date_class = class_df.index.min() + pd.Timedelta(days=training_days_class)
print(f"Cutoff date for training set Classic Bike Data: {cutoff_date_class}")

Cutoff date for training set Classic Bike Data: 2023-08-01 00:00:00


total_days_elect = (elect_df.index.max() - elect_df.index.min()).days + 1
print(f"Total number of days of Electric Bike Data: {total_days_elect}")

Total number of days of Electric Bike Data: 1357


training_days_elect = int(total_days_elect * 0.8)
cutoff_date_elect = elect_df.index.min() + pd.Timedelta(days=training_days_elect)
print(f"Cutoff date for training set Electric Bike Data: {cutoff_date_elect}")

Cutoff date for training set Electric Bike Data: 2023-07-03 00:00:00


# Split the data into training and test sets
class_train_df = class_df.loc[class_df.index < cutoff_date_class]
class_test_df = class_df.loc[class_df.index >= cutoff_date_class]

print(f"Shape of class_train_df: {class_train_df.shape}")
print(f"Shape of class_test_df: {class_test_df.shape}")

Shape of class_train_df: (971, 4)
Shape of class_test_df: (243, 4)


elect_train_df = elect_df.loc[elect_df.index < cutoff_date_elect]
elect_test_df = elect_df.loc[elect_df.index >= cutoff_date_elect]

print(f"Shape of elect_train_df: {elect_train_df.shape}")
print(f"Shape of celect_test_df: {elect_test_df.shape}")

Shape of elect_train_df: (1073, 5)
Shape of celect_test_df: (272, 5)


class_train_df.dropna(subset=['classic_bike_rentals'], inplace=True)
class_test_df.dropna(subset=['classic_bike_rentals'], inplace=True)

<ipython-input-39-c94de804c9af>:1: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  class_train_df.dropna(subset=['classic_bike_rentals'], inplace=True)
<ipython-input-39-c94de804c9af>:2: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  class_test_df.dropna(subset=['classic_bike_rentals'], inplace=True)


class_training_set = class_train_df['classic_bike_rentals'].values.reshape(-1, 1)
class_validation_set = class_test_df['classic_bike_rentals'].values.reshape(-1, 1)
# Combine training and validation sets
combined_set = np.concatenate((class_training_set, class_validation_set), axis=0)


# Fit the scaler to the training data
sc = MinMaxScaler(feature_range=(0, 1))

combined_set_scaled = sc.fit_transform(combined_set)

# Split the scaled data into training and validation sets
class_training_set_scaled = combined_set_scaled[:len(class_training_set), :]
class_validation_set_scaled = combined_set_scaled[len(class_training_set):, :]


def create_sequences_class(data, seq_length=30, n_lags=2):
    X = []
    y = []
    for i in range(seq_length + n_lags, len(data)):
        X.append(np.concatenate([data[i-seq_length-n_lags:i-n_lags, 0], data[i-n_lags:i, 0]], axis=0))
        y.append(data[i, 0])
    return np.array(X), np.array(y)

X_train_class, y_train_class = create_sequences_class(class_training_set_scaled)
X_validation_class, y_validation_class = create_sequences_class(class_validation_set_scaled)

# Reshape inputs for LSTM model
X_train_class = np.reshape(X_train_class, (X_train_class.shape[0], X_train_class.shape[1], 1))
X_validation_class = np.reshape(X_validation_class, (X_validation_class.shape[0], X_validation_class.shape[1], 1))


%%time

def build_model(hp):
    model = keras.Sequential()
    model.add(keras.layers.LSTM(
        units=hp.Int('units', min_value=32, max_value=512, step=32),
        return_sequences=True,
        input_shape=(X_train_class.shape[1], 1)
    ))
    model.add(keras.layers.Dropout(hp.Float('dropout_1', min_value=0.0, max_value=0.5, step=0.1)))
    for i in range(hp.Int('num_layers', 1, 5)):
        model.add(keras.layers.LSTM(
            units=hp.Int(f'lstm_{i}_units', min_value=32, max_value=512, step=32),
            return_sequences=True if i < hp.Int('num_layers', 1, 5) - 1 else False
        ))
        model.add(keras.layers.Dropout(hp.Float(f'dropout_{i+2}', min_value=0.0, max_value=0.5, step=0.1)))
    model.add(keras.layers.Dense(units=1))

    model.compile(
        optimizer=keras.optimizers.Adam(
            hp.Float('learning_rate', min_value=1e-5, max_value=1e-1, sampling='log')
        ),
        loss='mean_squared_error'
    )
    return model

tuner = kt.BayesianOptimization(
    build_model,
    objective='val_loss',
    max_trials=20,
    executions_per_trial=1,
    directory='my_dir',
    project_name='lstm_tuning',
    overwrite=True
)

tuner.search_space_summary()


tuner.search(
    X_train_class, y_train_class,
    epochs=100,
    batch_size=64,
    validation_data=(X_validation_class, y_validation_class),
    callbacks=[keras.callbacks.EarlyStopping(monitor='val_loss', patience=3)]
)

# Get the optimal hyperparameters
best_hps = tuner.get_best_hyperparameters(num_trials=1)[0]

# Build the model with the optimal hyperparameters and train it on the data
model_lstm_class = tuner.hypermodel.build(best_hps)

model_lstm_class.compile(optimizer='adam', loss='mean_squared_error')

history_lstm_class = model_lstm_class.fit(
    X_train_class, y_train_class,
    epochs=100, batch_size=64,
    validation_data=(X_validation_class, y_validation_class)
)

Trial 20 Complete [00h 03m 47s]
val_loss: 0.01859133690595627

Best val_loss So Far: 0.005138479173183441
Total elapsed time: 00h 45m 35s
Epoch 1/100
15/15 [==============================] - 14s 721ms/step - loss: 0.0334 - val_loss: 0.0060
Epoch 2/100
15/15 [==============================] - 8s 508ms/step - loss: 0.0123 - val_loss: 0.0054
Epoch 3/100
15/15 [==============================] - 10s 681ms/step - loss: 0.0111 - val_loss: 0.0055
Epoch 4/100
15/15 [==============================] - 8s 515ms/step - loss: 0.0116 - val_loss: 0.0054
Epoch 5/100
15/15 [==============================] - 10s 652ms/step - loss: 0.0114 - val_loss: 0.0071
Epoch 6/100
15/15 [==============================] - 9s 585ms/step - loss: 0.0107 - val_loss: 0.0062
Epoch 7/100
15/15 [==============================] - 9s 578ms/step - loss: 0.0109 - val_loss: 0.0060
Epoch 8/100
15/15 [==============================] - 10s 689ms/step - loss: 0.0111 - val_loss: 0.0052
Epoch 9/100
15/15 [==============================] - 8s 509ms/step - loss: 0.0113 - val_loss: 0.0052
Epoch 10/100
15/15 [==============================] - 10s 677ms/step - loss: 0.0117 - val_loss: 0.0065
Epoch 11/100
15/15 [==============================] - 8s 504ms/step - loss: 0.0109 - val_loss: 0.0056
Epoch 12/100
15/15 [==============================] - 10s 681ms/step - loss: 0.0109 - val_loss: 0.0052
Epoch 13/100
15/15 [==============================] - 9s 605ms/step - loss: 0.0110 - val_loss: 0.0059
Epoch 14/100
15/15 [==============================] - 9s 570ms/step - loss: 0.0110 - val_loss: 0.0051
Epoch 15/100
15/15 [==============================] - 10s 690ms/step - loss: 0.0109 - val_loss: 0.0055
Epoch 16/100
15/15 [==============================] - 8s 517ms/step - loss: 0.0107 - val_loss: 0.0051
Epoch 17/100
15/15 [==============================] - 10s 678ms/step - loss: 0.0112 - val_loss: 0.0062
Epoch 18/100
15/15 [==============================] - 8s 514ms/step - loss: 0.0115 - val_loss: 0.0051
Epoch 19/100
15/15 [==============================] - 12s 788ms/step - loss: 0.0105 - val_loss: 0.0050
Epoch 20/100
15/15 [==============================] - 11s 739ms/step - loss: 0.0105 - val_loss: 0.0053
Epoch 21/100
15/15 [==============================] - 8s 520ms/step - loss: 0.0105 - val_loss: 0.0056
Epoch 22/100
15/15 [==============================] - 10s 676ms/step - loss: 0.0107 - val_loss: 0.0052
Epoch 23/100
15/15 [==============================] - 8s 509ms/step - loss: 0.0104 - val_loss: 0.0054
Epoch 24/100
15/15 [==============================] - 10s 675ms/step - loss: 0.0101 - val_loss: 0.0075
Epoch 25/100
15/15 [==============================] - 8s 548ms/step - loss: 0.0115 - val_loss: 0.0050
Epoch 26/100
15/15 [==============================] - 9s 611ms/step - loss: 0.0113 - val_loss: 0.0059
Epoch 27/100
15/15 [==============================] - 10s 645ms/step - loss: 0.0106 - val_loss: 0.0049
Epoch 28/100
15/15 [==============================] - 8s 510ms/step - loss: 0.0099 - val_loss: 0.0059
Epoch 29/100
15/15 [==============================] - 10s 677ms/step - loss: 0.0106 - val_loss: 0.0052
Epoch 30/100
15/15 [==============================] - 8s 506ms/step - loss: 0.0106 - val_loss: 0.0049
Epoch 31/100
15/15 [==============================] - 10s 676ms/step - loss: 0.0100 - val_loss: 0.0049
Epoch 32/100
15/15 [==============================] - 8s 565ms/step - loss: 0.0100 - val_loss: 0.0049
Epoch 33/100
15/15 [==============================] - 9s 600ms/step - loss: 0.0099 - val_loss: 0.0056
Epoch 34/100
15/15 [==============================] - 9s 634ms/step - loss: 0.0098 - val_loss: 0.0052
Epoch 35/100
15/15 [==============================] - 8s 522ms/step - loss: 0.0100 - val_loss: 0.0049
Epoch 36/100
15/15 [==============================] - 10s 680ms/step - loss: 0.0099 - val_loss: 0.0053
Epoch 37/100
15/15 [==============================] - 8s 506ms/step - loss: 0.0102 - val_loss: 0.0049
Epoch 38/100
15/15 [==============================] - 10s 678ms/step - loss: 0.0105 - val_loss: 0.0047
Epoch 39/100
15/15 [==============================] - 8s 512ms/step - loss: 0.0100 - val_loss: 0.0049
Epoch 40/100
15/15 [==============================] - 10s 661ms/step - loss: 0.0102 - val_loss: 0.0055
Epoch 41/100
15/15 [==============================] - 8s 573ms/step - loss: 0.0103 - val_loss: 0.0048
Epoch 42/100
15/15 [==============================] - 9s 577ms/step - loss: 0.0105 - val_loss: 0.0047
Epoch 43/100
15/15 [==============================] - 10s 688ms/step - loss: 0.0093 - val_loss: 0.0047
Epoch 44/100
15/15 [==============================] - 8s 514ms/step - loss: 0.0093 - val_loss: 0.0047
Epoch 45/100
15/15 [==============================] - 10s 672ms/step - loss: 0.0099 - val_loss: 0.0047
Epoch 46/100
15/15 [==============================] - 8s 512ms/step - loss: 0.0093 - val_loss: 0.0050
Epoch 47/100
15/15 [==============================] - 12s 812ms/step - loss: 0.0098 - val_loss: 0.0054
Epoch 48/100
15/15 [==============================] - 10s 679ms/step - loss: 0.0095 - val_loss: 0.0046
Epoch 49/100
15/15 [==============================] - 8s 511ms/step - loss: 0.0098 - val_loss: 0.0046
Epoch 50/100
15/15 [==============================] - 10s 668ms/step - loss: 0.0099 - val_loss: 0.0046
Epoch 51/100
15/15 [==============================] - 8s 517ms/step - loss: 0.0091 - val_loss: 0.0046
Epoch 52/100
15/15 [==============================] - 10s 664ms/step - loss: 0.0094 - val_loss: 0.0047
Epoch 53/100
15/15 [==============================] - 9s 616ms/step - loss: 0.0095 - val_loss: 0.0048
Epoch 54/100
15/15 [==============================] - 8s 549ms/step - loss: 0.0092 - val_loss: 0.0047
Epoch 55/100
15/15 [==============================] - 10s 667ms/step - loss: 0.0092 - val_loss: 0.0051
Epoch 56/100
15/15 [==============================] - 8s 507ms/step - loss: 0.0095 - val_loss: 0.0045
Epoch 57/100
15/15 [==============================] - 10s 669ms/step - loss: 0.0095 - val_loss: 0.0045
Epoch 58/100
15/15 [==============================] - 7s 499ms/step - loss: 0.0091 - val_loss: 0.0045
Epoch 59/100
15/15 [==============================] - 10s 671ms/step - loss: 0.0091 - val_loss: 0.0046
Epoch 60/100
15/15 [==============================] - 8s 537ms/step - loss: 0.0094 - val_loss: 0.0046
Epoch 61/100
15/15 [==============================] - 9s 618ms/step - loss: 0.0096 - val_loss: 0.0045
Epoch 62/100
15/15 [==============================] - 9s 630ms/step - loss: 0.0093 - val_loss: 0.0045
Epoch 63/100
15/15 [==============================] - 8s 522ms/step - loss: 0.0092 - val_loss: 0.0048
Epoch 64/100
15/15 [==============================] - 10s 666ms/step - loss: 0.0092 - val_loss: 0.0045
Epoch 65/100
15/15 [==============================] - 8s 507ms/step - loss: 0.0095 - val_loss: 0.0051
Epoch 66/100
15/15 [==============================] - 10s 672ms/step - loss: 0.0093 - val_loss: 0.0053
Epoch 67/100
15/15 [==============================] - 8s 505ms/step - loss: 0.0096 - val_loss: 0.0050
Epoch 68/100
15/15 [==============================] - 10s 665ms/step - loss: 0.0091 - val_loss: 0.0046
Epoch 69/100
15/15 [==============================] - 9s 585ms/step - loss: 0.0090 - val_loss: 0.0047
Epoch 70/100
15/15 [==============================] - 9s 575ms/step - loss: 0.0090 - val_loss: 0.0054
Epoch 71/100
15/15 [==============================] - 10s 668ms/step - loss: 0.0090 - val_loss: 0.0046
Epoch 72/100
15/15 [==============================] - 8s 504ms/step - loss: 0.0090 - val_loss: 0.0045
Epoch 73/100
15/15 [==============================] - 10s 669ms/step - loss: 0.0094 - val_loss: 0.0050
Epoch 74/100
15/15 [==============================] - 8s 503ms/step - loss: 0.0093 - val_loss: 0.0045
Epoch 75/100
15/15 [==============================] - 12s 803ms/step - loss: 0.0092 - val_loss: 0.0047
Epoch 76/100
15/15 [==============================] - 10s 646ms/step - loss: 0.0096 - val_loss: 0.0045
Epoch 77/100
15/15 [==============================] - 8s 501ms/step - loss: 0.0094 - val_loss: 0.0057
Epoch 78/100
15/15 [==============================] - 10s 670ms/step - loss: 0.0093 - val_loss: 0.0051
Epoch 79/100
15/15 [==============================] - 8s 505ms/step - loss: 0.0094 - val_loss: 0.0046
Epoch 80/100
15/15 [==============================] - 10s 670ms/step - loss: 0.0097 - val_loss: 0.0045
Epoch 81/100
15/15 [==============================] - 8s 521ms/step - loss: 0.0095 - val_loss: 0.0044
Epoch 82/100
15/15 [==============================] - 10s 634ms/step - loss: 0.0091 - val_loss: 0.0044
Epoch 83/100
15/15 [==============================] - 9s 625ms/step - loss: 0.0092 - val_loss: 0.0045
Epoch 84/100
15/15 [==============================] - 8s 525ms/step - loss: 0.0091 - val_loss: 0.0046
Epoch 85/100
15/15 [==============================] - 10s 678ms/step - loss: 0.0094 - val_loss: 0.0045
Epoch 86/100
15/15 [==============================] - 8s 514ms/step - loss: 0.0089 - val_loss: 0.0046
Epoch 87/100
15/15 [==============================] - 10s 669ms/step - loss: 0.0091 - val_loss: 0.0046
Epoch 88/100
15/15 [==============================] - 7s 501ms/step - loss: 0.0091 - val_loss: 0.0048
Epoch 89/100
15/15 [==============================] - 10s 666ms/step - loss: 0.0090 - val_loss: 0.0047
Epoch 90/100
15/15 [==============================] - 8s 544ms/step - loss: 0.0090 - val_loss: 0.0047
Epoch 91/100
15/15 [==============================] - 9s 611ms/step - loss: 0.0089 - val_loss: 0.0046
Epoch 92/100
15/15 [==============================] - 9s 643ms/step - loss: 0.0089 - val_loss: 0.0047
Epoch 93/100
15/15 [==============================] - 8s 504ms/step - loss: 0.0085 - val_loss: 0.0046
Epoch 94/100
15/15 [==============================] - 10s 664ms/step - loss: 0.0087 - val_loss: 0.0049
Epoch 95/100
15/15 [==============================] - 7s 498ms/step - loss: 0.0087 - val_loss: 0.0045
Epoch 96/100
15/15 [==============================] - 10s 670ms/step - loss: 0.0086 - val_loss: 0.0044
Epoch 97/100
15/15 [==============================] - 8s 515ms/step - loss: 0.0088 - val_loss: 0.0045
Epoch 98/100
15/15 [==============================] - 10s 635ms/step - loss: 0.0086 - val_loss: 0.0048
Epoch 99/100
15/15 [==============================] - 9s 583ms/step - loss: 0.0088 - val_loss: 0.0049
Epoch 100/100
15/15 [==============================] - 9s 574ms/step - loss: 0.0086 - val_loss: 0.0044
CPU times: user 1h 21min 50s, sys: 1min 48s, total: 1h 23min 38s
Wall time: 1h 37s


# Get the training and validation loss histories
training_loss_class = history_lstm_class.history['loss']
validation_loss_class = history_lstm_class.history['val_loss']

# Create a plot for the training and validation loss
plt.figure(figsize=(10, 6))
plt.plot(training_loss_class, label='Training Loss')
plt.plot(validation_loss_class, label='Validation Loss')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.title('Training and Validation Loss (Classic Bike)')
plt.legend()
plt.show()


# Make predictions
y_pred_class = model_lstm_class.predict(X_validation_class)

# Invert scaling to compare with original prices
y_pred_rescaled_class = sc.inverse_transform(y_pred_class)
y_validation_rescaled_class = sc.inverse_transform(y_validation_class.reshape(-1, 1))

# Calculate the errors
mse_class = mean_squared_error(y_validation_rescaled_class, y_pred_rescaled_class)
rmse_class = np.sqrt(mse_class)
mae_class = mean_absolute_error(y_validation_rescaled_class, y_pred_rescaled_class)
mpe_class = np.mean((y_validation_rescaled_class - y_pred_rescaled_class) / y_validation_rescaled_class) * 100
mape_class = np.mean(np.abs((y_validation_rescaled_class - y_pred_rescaled_class) / y_validation_rescaled_class)) * 100

# Create a dataframe for the metrics
metrics_df_class = pd.DataFrame({
    'MSE': [mse_class],
    'RMSE': [rmse_class],
    'MAE': [mae_class],
    'MPE': [mpe_class],
    'MAPE': [mape_class]
})

print("\nLSTM Model of Evaluation (Classic Bike)\n")
metrics_df_class

7/7 [==============================] - 1s 88ms/step

LSTM Model of Evaluation (Classic Bike)


# Get the model residuals
val_loss_class = np.array(history_lstm_class.history['val_loss'])
loss_class = np.array(history_lstm_class.history['loss'])
residuals_class = val_loss_class - loss_class

# Plot the PACF
fig, ax = plt.subplots(figsize=(12, 6))
plot_pacf(residuals_class, lags=20, ax=ax)
plt.title('Partial Autocorrelation Function (PACF) of LSTM Classic Bike Residuals')
plt.show()


# Perform Ljung-Box test
lb_class_results = acorr_ljungbox(residuals_class, lags=[5, 10, 15, 20], return_df=True)
lb_class_results


# Scale the test data
class_test_df_scaled = sc.transform(class_test_df['classic_bike_rentals'].values.reshape(-1, 1))
X_test_class, y_test_class = create_sequences_class(class_test_df_scaled)
X_test_class = np.reshape(X_test_class, (X_test_class.shape[0], X_test_class.shape[1], 1))

y_pred_class = model_lstm_class.predict(X_test_class)

y_pred_class = sc.inverse_transform(y_pred_class)
y_test_class = sc.inverse_transform(y_test_class.reshape(-1, 1))

# Create a datetime index for the test data
test_dates_class = pd.date_range(start=class_test_df.index.min(), end=class_test_df.index.max(), freq='D')

# Get the length of y_test_class
y_test_len_class = y_test_class.shape[0]

# Slice test_dates to match y_test_class length
test_dates_class  = test_dates_class [-y_test_len_class:]


# Calculate the 95% confidence intervals
confidence_level_test_class = 0.95
residuals_class = y_test_class - y_pred_class
residual_std_class = np.std(residuals_class)
z_score_class = norm.ppf((1 + confidence_level_test_class) / 2)
lower_ci_class = y_pred_class - z_score_class * residual_std_class
upper_ci_class = y_pred_class + z_score_class * residual_std_class

# Flatten the confidence intervals to 1-dimensional arrays
lower_ci_class = lower_ci_class.ravel()
upper_ci_class = upper_ci_class.ravel()

# Create a datetime index for the training data
train_dates_class = pd.date_range(start=class_train_df.index.min(), end=class_train_df.index.max(), freq='D')

predicted_df_class = pd.DataFrame({
    'Start_Test_Date': test_dates_class,
    'Classic_Actual_Value': y_test_class.ravel(),
    'Classic_Predicted_Value': y_pred_class.ravel(),
    'Lower CI': lower_ci_class,
    'Upper CI': upper_ci_class
})

# Display the DataFrame
predicted_df_class.head()

7/7 [==============================] - 1s 190ms/step


# Save LSTMs Model prediction of classic bike result

predicted_df_class.to_csv("/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/predicted_df_classic_lstms.csv", index=False)


fig, ax = plt.subplots(figsize=(10, 6))
ax.plot(test_dates_class, y_test_class, label='Actual', marker='o', color = 'navy')
ax.plot(test_dates_class, y_pred_class, label='Predicted', marker = '.', color = 'darkorange')
ax.fill_between(test_dates_class, lower_ci_class, upper_ci_class, alpha=0.2, label='95% Confidence Interval', color='dimgray')
ax.set_xlabel('Date')
ax.set_ylabel('Classic Bike Rentals')
ax.set_title('Classic Bike Predictions with 95% Confidence Interval (LSTMs Model)')

plt.xticks(rotation=45)
ax.xaxis.set_major_formatter(mdates.DateFormatter('%Y-%m-%d'))
ax.legend()
plt.tight_layout()
plt.show()


fig, ax = plt.subplots(figsize=(12, 6))
ax.plot(train_dates_class, class_train_df['classic_bike_rentals'], label='Historical Data', alpha=0.8, color='darkgreen', linestyle='--')
ax.plot(test_dates_class, y_test_class, label='Actual', marker='o', color='navy', linestyle='-')
ax.plot(test_dates_class, y_pred_class, label='Predicted', marker='.', color='darkorange', linestyle=':')
ax.fill_between(test_dates_class, lower_ci_class, upper_ci_class, alpha=0.3, label='95% Confidence Interval', color='dimgray')
ax.set_xlabel('Date')
ax.set_ylabel('Classic Bike Rentals')
ax.set_title('Classic Bike Predictions(LSTMs Model)')
plt.xticks(rotation=45)
ax.xaxis.set_major_formatter(mdates.DateFormatter('%Y-%m-%d'))
ax.legend()
plt.tight_layout()
plt.show()


elect_train_df.dropna(subset=['elect_bike_rentals'], inplace=True)
elect_test_df.dropna(subset=['elect_bike_rentals'], inplace=True)

<ipython-input-63-4cb917406275>:1: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  elect_train_df.dropna(subset=['elect_bike_rentals'], inplace=True)
<ipython-input-63-4cb917406275>:2: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  elect_test_df.dropna(subset=['elect_bike_rentals'], inplace=True)


sc = MinMaxScaler(feature_range=(0, 1))

elect_training_set = elect_train_df['elect_bike_rentals'].values.reshape(-1, 1)
elect_validation_set = elect_test_df['elect_bike_rentals'].values.reshape(-1, 1)
elect_combined_set = np.concatenate((elect_training_set, elect_validation_set), axis=0)

# Fit the scaler to the training data
elect_combined_set_scaled = sc.fit_transform(elect_combined_set)

# Split the scaled data into training and validation sets
elect_training_set_scaled = elect_combined_set_scaled[:len(elect_training_set), :]
elect_validation_set_scaled = elect_combined_set_scaled[len(elect_training_set):, :]


def create_sequences_elect(data, seq_length=60, n_lags=2):
    X = []
    y = []
    for i in range(seq_length + n_lags, len(data)):
        X.append(np.concatenate([data[i-seq_length-n_lags:i-n_lags, 0], data[i-n_lags:i, 0]], axis=0))
        y.append(data[i, 0])
    return np.array(X), np.array(y)

X_train_elect, y_train_elect = create_sequences_elect(elect_training_set_scaled)
X_validation_elect, y_validation_elect = create_sequences_elect(elect_validation_set_scaled)

# Reshape inputs for LSTM model
X_train_elect = np.reshape(X_train_elect, (X_train_elect.shape[0], X_train_elect.shape[1], 1))
X_validation_elect = np.reshape(X_validation_elect, (X_validation_elect.shape[0], X_validation_elect.shape[1], 1))


%%time

def build_model(hp):
    model = keras.Sequential()
    model.add(keras.layers.LSTM(
        units=hp.Int('units', min_value=32, max_value=512, step=32),
        return_sequences=True,
        input_shape=(X_train_elect.shape[1], 1)
    ))
    model.add(keras.layers.Dropout(hp.Float('dropout_1', min_value=0.0, max_value=0.5, step=0.1)))
    for i in range(hp.Int('num_layers', 1, 5)):
        model.add(keras.layers.LSTM(
            units=hp.Int(f'lstm_{i}_units', min_value=32, max_value=512, step=32),
            return_sequences=True if i < hp.Int('num_layers', 1, 5) - 1 else False
        ))
        model.add(keras.layers.Dropout(hp.Float(f'dropout_{i+2}', min_value=0.0, max_value=0.5, step=0.1)))
    model.add(keras.layers.Dense(units=1))

    model.compile(
        optimizer=keras.optimizers.Adam(
            hp.Float('learning_rate', min_value=1e-5, max_value=1e-1, sampling='log')
        ),
        loss='mean_squared_error'
    )
    return model

tuner = kt.BayesianOptimization(
    build_model,
    objective='val_loss',
    max_trials=20,
    executions_per_trial=1,
    directory='my_dir',
    project_name='lstm_tuning',
    overwrite=True
)

tuner.search_space_summary()


tuner.search(
    X_train_elect, y_train_elect,
    epochs=100,
    batch_size=64,
    validation_data=(X_validation_elect, y_validation_elect),
    callbacks=[keras.callbacks.EarlyStopping(monitor='val_loss', patience=3)]
)

# Get the optimal hyperparameters
best_hps = tuner.get_best_hyperparameters(num_trials=1)[0]

# Build the model with the optimal hyperparameters and train it on the data
model_lstm_elect = tuner.hypermodel.build(best_hps)

model_lstm_elect.compile(optimizer = 'adam', loss = 'mean_squared_error')

history_lstm_elect = model_lstm_elect.fit(
    X_train_elect, y_train_elect,
    epochs=100, batch_size=64,
    validation_data=(X_validation_elect, y_validation_elect)
)

Trial 20 Complete [00h 01m 38s]
val_loss: 0.03181834891438484

Best val_loss So Far: 0.007478794548660517
Total elapsed time: 01h 11m 34s
Epoch 1/100
16/16 [==============================] - 22s 1s/step - loss: 0.0500 - val_loss: 0.0087
Epoch 2/100
16/16 [==============================] - 18s 1s/step - loss: 0.0125 - val_loss: 0.0079
Epoch 3/100
16/16 [==============================] - 18s 1s/step - loss: 0.0117 - val_loss: 0.0080
Epoch 4/100
16/16 [==============================] - 18s 1s/step - loss: 0.0109 - val_loss: 0.0094
Epoch 5/100
16/16 [==============================] - 18s 1s/step - loss: 0.0100 - val_loss: 0.0089
Epoch 6/100
16/16 [==============================] - 17s 1s/step - loss: 0.0106 - val_loss: 0.0077
Epoch 7/100
16/16 [==============================] - 18s 1s/step - loss: 0.0101 - val_loss: 0.0077
Epoch 8/100
16/16 [==============================] - 19s 1s/step - loss: 0.0099 - val_loss: 0.0083
Epoch 9/100
16/16 [==============================] - 21s 1s/step - loss: 0.0102 - val_loss: 0.0083
Epoch 10/100
16/16 [==============================] - 18s 1s/step - loss: 0.0103 - val_loss: 0.0077
Epoch 11/100
16/16 [==============================] - 20s 1s/step - loss: 0.0101 - val_loss: 0.0088
Epoch 12/100
16/16 [==============================] - 17s 1s/step - loss: 0.0097 - val_loss: 0.0076
Epoch 13/100
16/16 [==============================] - 18s 1s/step - loss: 0.0092 - val_loss: 0.0077
Epoch 14/100
16/16 [==============================] - 18s 1s/step - loss: 0.0094 - val_loss: 0.0079
Epoch 15/100
16/16 [==============================] - 17s 1s/step - loss: 0.0106 - val_loss: 0.0075
Epoch 16/100
16/16 [==============================] - 17s 1s/step - loss: 0.0095 - val_loss: 0.0076
Epoch 17/100
16/16 [==============================] - 17s 1s/step - loss: 0.0092 - val_loss: 0.0078
Epoch 18/100
16/16 [==============================] - 18s 1s/step - loss: 0.0097 - val_loss: 0.0082
Epoch 19/100
16/16 [==============================] - 18s 1s/step - loss: 0.0103 - val_loss: 0.0074
Epoch 20/100
16/16 [==============================] - 17s 1s/step - loss: 0.0098 - val_loss: 0.0083
Epoch 21/100
16/16 [==============================] - 17s 1s/step - loss: 0.0097 - val_loss: 0.0081
Epoch 22/100
16/16 [==============================] - 17s 1s/step - loss: 0.0104 - val_loss: 0.0096
Epoch 23/100
16/16 [==============================] - 17s 1s/step - loss: 0.0100 - val_loss: 0.0080
Epoch 24/100
16/16 [==============================] - 21s 1s/step - loss: 0.0094 - val_loss: 0.0076
Epoch 25/100
16/16 [==============================] - 18s 1s/step - loss: 0.0093 - val_loss: 0.0075
Epoch 26/100
16/16 [==============================] - 17s 1s/step - loss: 0.0096 - val_loss: 0.0073
Epoch 27/100
16/16 [==============================] - 17s 1s/step - loss: 0.0093 - val_loss: 0.0082
Epoch 28/100
16/16 [==============================] - 17s 1s/step - loss: 0.0095 - val_loss: 0.0077
Epoch 29/100
16/16 [==============================] - 17s 1s/step - loss: 0.0092 - val_loss: 0.0073
Epoch 30/100
16/16 [==============================] - 17s 1s/step - loss: 0.0097 - val_loss: 0.0072
Epoch 31/100
16/16 [==============================] - 19s 1s/step - loss: 0.0098 - val_loss: 0.0074
Epoch 32/100
16/16 [==============================] - 18s 1s/step - loss: 0.0095 - val_loss: 0.0073
Epoch 33/100
16/16 [==============================] - 17s 1s/step - loss: 0.0097 - val_loss: 0.0077
Epoch 34/100
16/16 [==============================] - 17s 1s/step - loss: 0.0097 - val_loss: 0.0079
Epoch 35/100
16/16 [==============================] - 18s 1s/step - loss: 0.0103 - val_loss: 0.0087
Epoch 36/100
16/16 [==============================] - 18s 1s/step - loss: 0.0096 - val_loss: 0.0073
Epoch 37/100
16/16 [==============================] - 17s 1s/step - loss: 0.0093 - val_loss: 0.0088
Epoch 38/100
16/16 [==============================] - 20s 1s/step - loss: 0.0096 - val_loss: 0.0070
Epoch 39/100
16/16 [==============================] - 17s 1s/step - loss: 0.0095 - val_loss: 0.0070
Epoch 40/100
16/16 [==============================] - 20s 1s/step - loss: 0.0091 - val_loss: 0.0076
Epoch 41/100
16/16 [==============================] - 17s 1s/step - loss: 0.0092 - val_loss: 0.0069
Epoch 42/100
16/16 [==============================] - 17s 1s/step - loss: 0.0089 - val_loss: 0.0070
Epoch 43/100
16/16 [==============================] - 18s 1s/step - loss: 0.0088 - val_loss: 0.0069
Epoch 44/100
16/16 [==============================] - 20s 1s/step - loss: 0.0096 - val_loss: 0.0070
Epoch 45/100
16/16 [==============================] - 18s 1s/step - loss: 0.0091 - val_loss: 0.0070
Epoch 46/100
16/16 [==============================] - 17s 1s/step - loss: 0.0090 - val_loss: 0.0072
Epoch 47/100
16/16 [==============================] - 18s 1s/step - loss: 0.0087 - val_loss: 0.0077
Epoch 48/100
16/16 [==============================] - 17s 1s/step - loss: 0.0094 - val_loss: 0.0073
Epoch 49/100
16/16 [==============================] - 18s 1s/step - loss: 0.0090 - val_loss: 0.0067
Epoch 50/100
16/16 [==============================] - 19s 1s/step - loss: 0.0089 - val_loss: 0.0067
Epoch 51/100
16/16 [==============================] - 18s 1s/step - loss: 0.0090 - val_loss: 0.0069
Epoch 52/100
16/16 [==============================] - 18s 1s/step - loss: 0.0089 - val_loss: 0.0070
Epoch 53/100
16/16 [==============================] - 17s 1s/step - loss: 0.0090 - val_loss: 0.0066
Epoch 54/100
16/16 [==============================] - 17s 1s/step - loss: 0.0084 - val_loss: 0.0067
Epoch 55/100
16/16 [==============================] - 21s 1s/step - loss: 0.0093 - val_loss: 0.0066
Epoch 56/100
16/16 [==============================] - 19s 1s/step - loss: 0.0092 - val_loss: 0.0067
Epoch 57/100
16/16 [==============================] - 17s 1s/step - loss: 0.0091 - val_loss: 0.0071
Epoch 58/100
16/16 [==============================] - 18s 1s/step - loss: 0.0088 - val_loss: 0.0067
Epoch 59/100
16/16 [==============================] - 17s 1s/step - loss: 0.0083 - val_loss: 0.0065
Epoch 60/100
16/16 [==============================] - 17s 1s/step - loss: 0.0085 - val_loss: 0.0067
Epoch 61/100
16/16 [==============================] - 17s 1s/step - loss: 0.0090 - val_loss: 0.0066
Epoch 62/100
16/16 [==============================] - 18s 1s/step - loss: 0.0086 - val_loss: 0.0064
Epoch 63/100
16/16 [==============================] - 19s 1s/step - loss: 0.0088 - val_loss: 0.0065
Epoch 64/100
16/16 [==============================] - 17s 1s/step - loss: 0.0081 - val_loss: 0.0064
Epoch 65/100
16/16 [==============================] - 17s 1s/step - loss: 0.0083 - val_loss: 0.0066
Epoch 66/100
16/16 [==============================] - 17s 1s/step - loss: 0.0087 - val_loss: 0.0067
Epoch 67/100
16/16 [==============================] - 17s 1s/step - loss: 0.0084 - val_loss: 0.0064
Epoch 68/100
16/16 [==============================] - 17s 1s/step - loss: 0.0084 - val_loss: 0.0065
Epoch 69/100
16/16 [==============================] - 19s 1s/step - loss: 0.0086 - val_loss: 0.0067
Epoch 70/100
16/16 [==============================] - 17s 1s/step - loss: 0.0087 - val_loss: 0.0063
Epoch 71/100
16/16 [==============================] - 20s 1s/step - loss: 0.0083 - val_loss: 0.0064
Epoch 72/100
16/16 [==============================] - 17s 1s/step - loss: 0.0084 - val_loss: 0.0066
Epoch 73/100
16/16 [==============================] - 17s 1s/step - loss: 0.0084 - val_loss: 0.0064
Epoch 74/100
16/16 [==============================] - 17s 1s/step - loss: 0.0084 - val_loss: 0.0076
Epoch 75/100
16/16 [==============================] - 19s 1s/step - loss: 0.0086 - val_loss: 0.0062
Epoch 76/100
16/16 [==============================] - 18s 1s/step - loss: 0.0082 - val_loss: 0.0064
Epoch 77/100
16/16 [==============================] - 17s 1s/step - loss: 0.0083 - val_loss: 0.0069
Epoch 78/100
16/16 [==============================] - 17s 1s/step - loss: 0.0085 - val_loss: 0.0062
Epoch 79/100
16/16 [==============================] - 17s 1s/step - loss: 0.0082 - val_loss: 0.0062
Epoch 80/100
16/16 [==============================] - 17s 1s/step - loss: 0.0085 - val_loss: 0.0071
Epoch 81/100
16/16 [==============================] - 17s 1s/step - loss: 0.0084 - val_loss: 0.0064
Epoch 82/100
16/16 [==============================] - 19s 1s/step - loss: 0.0082 - val_loss: 0.0063
Epoch 83/100
16/16 [==============================] - 18s 1s/step - loss: 0.0085 - val_loss: 0.0067
Epoch 84/100
16/16 [==============================] - 17s 1s/step - loss: 0.0082 - val_loss: 0.0063
Epoch 85/100
16/16 [==============================] - 17s 1s/step - loss: 0.0088 - val_loss: 0.0066
Epoch 86/100
16/16 [==============================] - 17s 1s/step - loss: 0.0088 - val_loss: 0.0064
Epoch 87/100
16/16 [==============================] - 20s 1s/step - loss: 0.0079 - val_loss: 0.0064
Epoch 88/100
16/16 [==============================] - 17s 1s/step - loss: 0.0082 - val_loss: 0.0061
Epoch 89/100
16/16 [==============================] - 17s 1s/step - loss: 0.0085 - val_loss: 0.0062
Epoch 90/100
16/16 [==============================] - 17s 1s/step - loss: 0.0090 - val_loss: 0.0070
Epoch 91/100
16/16 [==============================] - 18s 1s/step - loss: 0.0086 - val_loss: 0.0061
Epoch 92/100
16/16 [==============================] - 18s 1s/step - loss: 0.0077 - val_loss: 0.0063
Epoch 93/100
16/16 [==============================] - 17s 1s/step - loss: 0.0084 - val_loss: 0.0060
Epoch 94/100
16/16 [==============================] - 17s 1s/step - loss: 0.0080 - val_loss: 0.0061
Epoch 95/100
16/16 [==============================] - 17s 1s/step - loss: 0.0080 - val_loss: 0.0060
Epoch 96/100
16/16 [==============================] - 17s 1s/step - loss: 0.0086 - val_loss: 0.0084
Epoch 97/100
16/16 [==============================] - 17s 1s/step - loss: 0.0084 - val_loss: 0.0060
Epoch 98/100
16/16 [==============================] - 19s 1s/step - loss: 0.0083 - val_loss: 0.0062
Epoch 99/100
16/16 [==============================] - 18s 1s/step - loss: 0.0078 - val_loss: 0.0059
Epoch 100/100
16/16 [==============================] - 17s 1s/step - loss: 0.0079 - val_loss: 0.0063
CPU times: user 2h 21min 27s, sys: 3min 18s, total: 2h 24min 45s
Wall time: 1h 42min 4s


print(model_lstm_elect.summary())

Model: "sequential_1"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 lstm_2 (LSTM)               (None, 62, 448)           806400    
                                                                 
 dropout_2 (Dropout)         (None, 62, 448)           0         
                                                                 
 lstm_3 (LSTM)               (None, 192)               492288    
                                                                 
 dropout_3 (Dropout)         (None, 192)               0         
                                                                 
 dense_1 (Dense)             (None, 1)                 193       
                                                                 
=================================================================
Total params: 1298881 (4.95 MB)
Trainable params: 1298881 (4.95 MB)
Non-trainable params: 0 (0.00 Byte)
_________________________________________________________________
None


# Get the training and validation loss histories
training_loss_elect = history_lstm_elect.history['loss']
validation_loss_elect = history_lstm_elect.history['val_loss']

# Create a plot for the training and validation loss
plt.figure(figsize=(10, 6))
plt.plot(training_loss_elect, label='Training Loss')
plt.plot(validation_loss_elect, label='Validation Loss')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.title('Training and Validation Loss (Electric Bike)')
plt.legend()
plt.show()


# Make predictions
y_pred_elect = model_lstm_elect.predict(X_validation_elect)

# Invert scaling to compare with original prices
y_pred_rescaled_elect = sc.inverse_transform(y_pred_elect)
y_validation_rescaled_elect = sc.inverse_transform(y_validation_elect.reshape(-1, 1))

# Calculate the errors
mse_elect = mean_squared_error(y_validation_rescaled_elect, y_pred_rescaled_elect)
rmse_elect = np.sqrt(mean_squared_error(y_validation_rescaled_elect, y_pred_rescaled_elect))
mae_elect = mean_absolute_error(y_validation_rescaled_elect, y_pred_rescaled_elect)
mpe_elect = np.mean((y_validation_rescaled_elect - y_pred_rescaled_elect) / y_validation_rescaled_elect) * 100
mape_elect = np.mean(np.abs((y_validation_rescaled_elect - y_pred_rescaled_elect) / y_validation_rescaled_elect)) * 100


# Create a dataframe for the metrics
metrics_df_elect = pd.DataFrame({
    'MSE': [mse_elect],
    'RMSE': [rmse_elect],
    'MAE': [mae_elect],
    'MPE': [mpe_elect],
    'MAPE': [mape_elect]
})

print("\nLSTM Model of Evaluation (Electric Bike)\n")
metrics_df_elect

7/7 [==============================] - 2s 166ms/step

LSTM Model of Evaluation (Electric Bike)


# Get the model residuals
val_loss_elect = np.array(history_lstm_elect.history['val_loss'])
loss_elect = np.array(history_lstm_elect.history['loss'])
residuals_elect = val_loss_elect - loss_elect

# Plot the PACF
fig, ax = plt.subplots(figsize=(12, 6))
plot_pacf(residuals_elect, lags=20, ax=ax)
plt.title('Partial Autocorrelation Function (PACF) of LSTM Electric Bike Residuals')
plt.show()


# Perform Ljung-Box test
lb_elect_results = acorr_ljungbox(residuals_elect, lags=[5, 10, 15, 20], return_df=True)
lb_elect_results


# Scale the test data
elect_test_df_scaled = sc.transform(elect_test_df['elect_bike_rentals'].values.reshape(-1, 1))
X_test_elect, y_test_elect = create_sequences_class(elect_test_df_scaled)
X_test_elect = np.reshape(X_test_elect, (X_test_elect.shape[0], X_test_elect.shape[1], 1))

y_pred_elect = model_lstm_elect.predict(X_test_elect)

y_pred_elect = sc.inverse_transform(y_pred_elect)
y_test_elect = sc.inverse_transform(y_test_elect.reshape(-1, 1))

# Create a datetime index for the test data
test_dates_elect = pd.date_range(start=elect_test_df.index.min(), end=elect_test_df.index.max(), freq='D')

# Get the length of y_test_class
y_test_len_elect = y_test_elect.shape[0]

# Slice test_dates to match y_test_class length
test_dates_elect  = test_dates_elect[-y_test_len_elect:]


# Calculate the 95% confidence intervals
confidence_level_test_elect = 0.95
residuals_elect = y_test_elect - y_pred_elect
residual_std_elect = np.std(residuals_elect)
z_score_elect = norm.ppf((1 + confidence_level_test_elect) / 2)
lower_ci_elect = y_pred_elect - z_score_elect * residual_std_elect
upper_ci_elect = y_pred_elect + z_score_elect * residual_std_elect

# Flatten the confidence intervals to 1-dimensional arrays
lower_ci_elect = lower_ci_elect.ravel()
upper_ci_elect = upper_ci_elect.ravel()

# Create a datetime index for the training data
train_dates_elect = pd.date_range(start=elect_train_df.index.min(), end=elect_train_df.index.max(), freq='D')

predicted_df_elect = pd.DataFrame({
    'Start_Test_Date': test_dates_elect,
    'Electric_Actual_Value': y_test_elect.ravel(),
    'Electric_Predicted_Value': y_pred_elect.ravel(),
    'Lower CI': lower_ci_elect,
    'Upper CI': upper_ci_elect
})

# Display the DataFrame
predicted_df_elect.head()

8/8 [==============================] - 2s 147ms/step


# Save LSTMs Model prediction of electric bike result

predicted_df_elect.to_csv("/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/predicted_df_electric_lstms.csv", index=True)


fig, ax = plt.subplots(figsize=(12, 6))
ax.plot(test_dates_elect, y_test_elect, label='Actual', marker='o', color = 'purple')
ax.plot(test_dates_elect, y_pred_elect, label='Predicted', marker = '.', color='orange')
ax.fill_between(test_dates_elect, lower_ci_elect, upper_ci_elect, alpha=0.2, label='95% Confidence Interval', color='dimgray')
ax.set_xlabel('Date')
ax.set_ylabel('Electric Bike Rentals')
ax.set_title('Electric Bike Predictions with 95% Confidence Interval (LSTMs Model)')
plt.xticks(rotation=45)
ax.xaxis.set_major_formatter(mdates.DateFormatter('%Y-%m-%d'))
ax.legend()
plt.tight_layout()
plt.show()


fig, ax = plt.subplots(figsize=(12, 6))
shorter_length = len(elect_train_df['elect_bike_rentals'])
train_dates_elect = train_dates_elect[:shorter_length]

ax.plot(train_dates_elect, elect_train_df['elect_bike_rentals'], label='Historical Data', alpha=0.5, color='green', linestyle='--')
ax.plot(test_dates_elect, y_test_elect, label='Actual', marker='o', color = 'purple', linestyle='-')
ax.plot(test_dates_elect, y_pred_elect, label='Predicted', marker='.', color='orange', linestyle=':')
ax.fill_between(test_dates_elect, lower_ci_elect, upper_ci_elect, alpha=0.2, label='95% Confidence Interval', color='gray')
ax.set_xlabel('Date')
ax.set_ylabel('Electric Bike Rentals')
ax.set_title('Electric Bike Predictions with 95% Confidence Interval (LSTMs Model)')
plt.xticks(rotation=45)
ax.xaxis.set_major_formatter(mdates.DateFormatter('%Y-%m-%d'))
ax.legend()
plt.tight_layout()
plt.show()


# identify potential parameters by trying ACF and PACF
fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(12, 8))

# Plot the Autocorrelation Function (ACF)
plot_acf(class_train_df['Differenced'].dropna(), lags=40, ax=ax1)

# Plot the Partial Autocorrelation Function (PACF)
plot_pacf(class_train_df['Differenced'].dropna(), lags=40, ax=ax2)

plt.show()


model = auto_arima(class_train_df[class_train_df.columns[-1]], start_p=1, start_q=1,
                   test='adf',       # use adftest to find optimal 'd'
                   max_p=3, max_q=3, # maximum p and q
                   m=1,              # frequency of series
                   d=0,           # we already difference the data
                   seasonal=False,   # No Seasonality
                   start_P=0,
                   D=0,
                   trace=True,
                   error_action='ignore',
                   suppress_warnings=True,
                   stepwise=True)

print(model.summary())

Performing stepwise search to minimize aic
 ARIMA(1,0,1)(0,0,0)[0]             : AIC=823.559, Time=0.17 sec
 ARIMA(0,0,0)(0,0,0)[0]             : AIC=998.278, Time=0.08 sec
 ARIMA(1,0,0)(0,0,0)[0]             : AIC=946.927, Time=0.05 sec
 ARIMA(0,0,1)(0,0,0)[0]             : AIC=891.630, Time=0.08 sec
 ARIMA(2,0,1)(0,0,0)[0]             : AIC=822.298, Time=0.27 sec
 ARIMA(2,0,0)(0,0,0)[0]             : AIC=893.406, Time=0.12 sec
 ARIMA(3,0,1)(0,0,0)[0]             : AIC=824.204, Time=0.29 sec
 ARIMA(2,0,2)(0,0,0)[0]             : AIC=821.903, Time=0.38 sec
 ARIMA(1,0,2)(0,0,0)[0]             : AIC=822.128, Time=0.23 sec
 ARIMA(3,0,2)(0,0,0)[0]             : AIC=821.853, Time=0.93 sec
 ARIMA(3,0,3)(0,0,0)[0]             : AIC=823.840, Time=1.38 sec
 ARIMA(2,0,3)(0,0,0)[0]             : AIC=821.899, Time=1.13 sec
 ARIMA(3,0,2)(0,0,0)[0] intercept   : AIC=823.413, Time=3.07 sec

Best model:  ARIMA(3,0,2)(0,0,0)[0]          
Total fit time: 8.196 seconds
                               SARIMAX Results                                
==============================================================================
Dep. Variable:                      y   No. Observations:                  971
Model:               SARIMAX(3, 0, 2)   Log Likelihood                -404.927
Date:                Mon, 13 May 2024   AIC                            821.853
Time:                        22:24:56   BIC                            851.123
Sample:                    12-03-2020   HQIC                           832.994
                         - 07-31-2023                                         
Covariance Type:                  opg                                         
==============================================================================
                 coef    std err          z      P>|z|      [0.025      0.975]
------------------------------------------------------------------------------
ar.L1         -0.5529      0.042    -13.012      0.000      -0.636      -0.470
ar.L2          0.3441      0.054      6.318      0.000       0.237       0.451
ar.L3         -0.0542      0.035     -1.541      0.123      -0.123       0.015
ma.L1          0.1678      0.035      4.764      0.000       0.099       0.237
ma.L2         -0.8049      0.031    -26.224      0.000      -0.865      -0.745
sigma2         0.1347      0.004     33.424      0.000       0.127       0.143
===================================================================================
Ljung-Box (L1) (Q):                   0.03   Jarque-Bera (JB):               421.16
Prob(Q):                              0.86   Prob(JB):                         0.00
Heteroskedasticity (H):               0.95   Skew:                            -0.69
Prob(H) (two-sided):                  0.65   Kurtosis:                         5.92
===================================================================================

Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).


model_selected_AIC = ARIMA(class_train_df[class_train_df.columns[-1]], order=(3, 0, 2))
fitted_model_AIC = model_selected_AIC.fit()

# Print out the summary of the model
print(fitted_model_AIC.summary())

/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/base/tsa_model.py:473: ValueWarning: No frequency information was provided, so inferred frequency D will be used.
  self._init_dates(dates, freq)
/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/base/tsa_model.py:473: ValueWarning: No frequency information was provided, so inferred frequency D will be used.
  self._init_dates(dates, freq)
/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/base/tsa_model.py:473: ValueWarning: No frequency information was provided, so inferred frequency D will be used.
  self._init_dates(dates, freq)

                               SARIMAX Results                                
==============================================================================
Dep. Variable:            Differenced   No. Observations:                  971
Model:                 ARIMA(3, 0, 2)   Log Likelihood                -404.686
Date:                Mon, 13 May 2024   AIC                            823.371
Time:                        22:25:00   BIC                            857.519
Sample:                    12-03-2020   HQIC                           836.368
                         - 07-31-2023                                         
Covariance Type:                  opg                                         
==============================================================================
                 coef    std err          z      P>|z|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0023      0.004      0.637      0.524      -0.005       0.010
ar.L1         -0.5520      0.043    -12.918      0.000      -0.636      -0.468
ar.L2          0.3453      0.054      6.342      0.000       0.239       0.452
ar.L3         -0.0541      0.035     -1.532      0.126      -0.123       0.015
ma.L1          0.1666      0.035      4.743      0.000       0.098       0.235
ma.L2         -0.8062      0.031    -26.350      0.000      -0.866      -0.746
sigma2         0.1347      0.004     31.797      0.000       0.126       0.143
===================================================================================
Ljung-Box (L1) (Q):                   0.04   Jarque-Bera (JB):               421.68
Prob(Q):                              0.84   Prob(JB):                         0.00
Heteroskedasticity (H):               0.95   Skew:                            -0.69
Prob(H) (two-sided):                  0.67   Kurtosis:                         5.92
===================================================================================

Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).

/usr/local/lib/python3.10/dist-packages/statsmodels/base/model.py:607: ConvergenceWarning: Maximum Likelihood optimization failed to converge. Check mle_retvals
  warnings.warn("Maximum Likelihood optimization failed to "


model_selected_BIC = ARIMA(class_train_df[class_train_df.columns[-1]], order=(1, 0, 1))
fitted_model_bic = model_selected_BIC.fit()

# Print out the summary of the model
print(fitted_model_bic.summary())

/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/base/tsa_model.py:473: ValueWarning: No frequency information was provided, so inferred frequency D will be used.
  self._init_dates(dates, freq)
/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/base/tsa_model.py:473: ValueWarning: No frequency information was provided, so inferred frequency D will be used.
  self._init_dates(dates, freq)
/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/base/tsa_model.py:473: ValueWarning: No frequency information was provided, so inferred frequency D will be used.
  self._init_dates(dates, freq)

                               SARIMAX Results                                
==============================================================================
Dep. Variable:            Differenced   No. Observations:                  971
Model:                 ARIMA(1, 0, 1)   Log Likelihood                -408.515
Date:                Mon, 13 May 2024   AIC                            825.031
Time:                        22:25:01   BIC                            844.544
Sample:                    12-03-2020   HQIC                           832.458
                         - 07-31-2023                                         
Covariance Type:                  opg                                         
==============================================================================
                 coef    std err          z      P>|z|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.0023      0.003      0.672      0.502      -0.004       0.009
ar.L1          0.4388      0.036     12.045      0.000       0.367       0.510
ma.L1         -0.8500      0.023    -37.237      0.000      -0.895      -0.805
sigma2         0.1357      0.004     32.687      0.000       0.128       0.144
===================================================================================
Ljung-Box (L1) (Q):                   0.71   Jarque-Bera (JB):               459.82
Prob(Q):                              0.40   Prob(JB):                         0.00
Heteroskedasticity (H):               0.94   Skew:                            -0.71
Prob(H) (two-sided):                  0.58   Kurtosis:                         6.06
===================================================================================

Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).


# Try to add seasonality
# First try weekly sedasonality
# Fitting a SARIMA model with suspected weekly seasonality
s = 7  # Weekly seasonality
model_sarima_week = SARIMAX(class_train_df['Differenced'],
                       order=(1, 0, 1),
                       seasonal_order=(1, 0, 1, s))

fitted_model_sarima_week = model_sarima_week.fit()

# Summary of the fitted SARIMA model
print(fitted_model_sarima_week.summary())

/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/base/tsa_model.py:473: ValueWarning: No frequency information was provided, so inferred frequency D will be used.
  self._init_dates(dates, freq)
/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/base/tsa_model.py:473: ValueWarning: No frequency information was provided, so inferred frequency D will be used.
  self._init_dates(dates, freq)

                                     SARIMAX Results                                     
=========================================================================================
Dep. Variable:                       Differenced   No. Observations:                  971
Model:             SARIMAX(1, 0, 1)x(1, 0, 1, 7)   Log Likelihood                -396.977
Date:                           Mon, 13 May 2024   AIC                            803.955
Time:                                   22:25:02   BIC                            828.346
Sample:                               12-03-2020   HQIC                           813.238
                                    - 07-31-2023                                         
Covariance Type:                             opg                                         
==============================================================================
                 coef    std err          z      P>|z|      [0.025      0.975]
------------------------------------------------------------------------------
ar.L1          0.4582      0.035     13.122      0.000       0.390       0.527
ma.L1         -0.8663      0.022    -39.660      0.000      -0.909      -0.824
ar.S.L7        0.9015      0.054     16.647      0.000       0.795       1.008
ma.S.L7       -0.8333      0.068    -12.306      0.000      -0.966      -0.701
sigma2         0.1325      0.004     35.223      0.000       0.125       0.140
===================================================================================
Ljung-Box (L1) (Q):                   0.23   Jarque-Bera (JB):               497.85
Prob(Q):                              0.63   Prob(JB):                         0.00
Heteroskedasticity (H):               0.93   Skew:                            -0.72
Prob(H) (two-sided):                  0.51   Kurtosis:                         6.20
===================================================================================

Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).


residuals = fitted_model_sarima_week.resid
fig, ax = plt.subplots(2, 2, figsize=(14, 8))

# Time series plot of residuals
ax[0, 0].plot(residuals, color='blue')
ax[0, 0].set_title('Residuals')
ax[0, 0].set_xlabel('Date')
ax[0, 0].set_ylabel('Residual')

# Histogram of residuals
ax[0, 1].hist(residuals, bins=30, color='blue', alpha=0.7)
ax[0, 1].set_title('Histogram of Residuals of Weekly SARIMA')

# Q-Q plot of residuals to check normality
sm.qqplot(residuals, line='s', ax=ax[1, 0])
ax[1, 0].set_title('Q-Q Plot of Residuals of Weekly SARIMA')

# ACF plot of residuals to check autocorrelations
sm.graphics.tsa.plot_acf(residuals, lags=40, ax=ax[1, 1])
ax[1, 1].set_title('Autocorrelation Function of Weekly SARIMA')

plt.tight_layout()
plt.show()


forecast = fitted_model_sarima_week.get_forecast(steps=len(class_test_df))
forecast_mean = forecast.predicted_mean

# Retrieve the last log value from the training set
last_log_value = class_train_df['log_classic_bike_rentals'].iloc[-1]

# Cumulative sum of the forecasted differenced values added to the last known log value
cumulative_forecast = last_log_value + forecast_mean.cumsum()

# Exponentiate to reverse the log transformation
original_scale_forecast = np.exp(cumulative_forecast)

# Actual values in original scale for comparison
actual_values = np.exp(class_test_df['log_classic_bike_rentals'])

# Calculate RMSE and MAE on the original scale
rmse_original = np.sqrt(mean_squared_error(actual_values, original_scale_forecast))
mae_original = mean_absolute_error(actual_values, original_scale_forecast)

original_scale_forecast.head(), rmse_original, mae_original

(2023-08-01    12440.132406
 2023-08-02    12147.910183
 2023-08-03    12921.383570
 2023-08-04    12845.423973
 2023-08-05    13715.486584
 Freq: D, Name: predicted_mean, dtype: float64,
 9031.539660169708,
 7949.9743306453565)


# Create a dataframe for the metrics
sarima_metrics_df_class = pd.DataFrame({
    'RMSE': [rmse_original],
    'MAE': [mae_original]
})

print("\nSARIMA Model of Evaluation (Classic Bike)\n")
sarima_metrics_df_class

SARIMA Model of Evaluation (Classic Bike)


plt.figure(figsize=(12, 6))
plt.plot(actual_values.index, actual_values, label='Actual', color='blue')
plt.plot(original_scale_forecast.index, original_scale_forecast, label='Forecast', color='red', linestyle='--')
plt.title('Comparison of Actual and Forecasted Classic Bike Rentals with Weekly SARIMA')
plt.xlabel('Date')
plt.ylabel('Number of Rentals')
plt.legend()
plt.grid(True)
plt.show()


# Try to add seasonality
# Then try monthly sedasonality
s = 30
model_sarima_mo = SARIMAX(class_train_df['Differenced'],
                       order=(1, 0, 1),
                       seasonal_order=(1, 0, 1, s))

fitted_model_sarima_mo = model_sarima_week.fit()

# Summary of the fitted SARIMA model
fitted_model_sarima_mo.summary()

/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/base/tsa_model.py:473: ValueWarning: No frequency information was provided, so inferred frequency D will be used.
  self._init_dates(dates, freq)
/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/base/tsa_model.py:473: ValueWarning: No frequency information was provided, so inferred frequency D will be used.
  self._init_dates(dates, freq)


residuals = fitted_model_sarima_mo.resid

fig, ax = plt.subplots(2, 2, figsize=(14, 8))

# Time series plot of residuals
ax[0, 0].plot(residuals, color='blue')
ax[0, 0].set_title('Residuals')
ax[0, 0].set_xlabel('Date')
ax[0, 0].set_ylabel('Residual')

# Histogram of residuals
ax[0, 1].hist(residuals, bins=30, color='blue', alpha=0.7)
ax[0, 1].set_title('Histogram of Residuals of Monthly SARIMA')

# Q-Q plot of residuals to check normality
sm.qqplot(residuals, line='s', ax=ax[1, 0])
ax[1, 0].set_title('Q-Q Plot of Residuals of Monthly SARIMA')

# ACF plot of residuals to check autocorrelations
sm.graphics.tsa.plot_acf(residuals, lags=40, ax=ax[1, 1])
ax[1, 1].set_title('Autocorrelation Function of Weekly SARIMA')

plt.tight_layout()
plt.show()


forecast = fitted_model_sarima_mo.get_forecast(steps=len(class_test_df))
forecast_mean = forecast.predicted_mean

# Retrieve the last log value from the training set
last_log_value = class_train_df['log_classic_bike_rentals'].iloc[-1]

# Cumulative sum of the forecasted differenced values added to the last known log value
cumulative_forecast = last_log_value + forecast_mean.cumsum()

# Exponentiate to reverse the log transformation
original_scale_forecast = np.exp(cumulative_forecast)

# Actual values in original scale for comparison

actual_values=class_test_df['classic_bike_rentals']
# Calculate RMSE and MAE on the original scale
rmse_original = np.sqrt(mean_squared_error(actual_values, original_scale_forecast))
mae_original = mean_absolute_error(actual_values, original_scale_forecast)

original_scale_forecast.head(), rmse_original, mae_original

(2023-08-01    12440.132406
 2023-08-02    12147.910183
 2023-08-03    12921.383570
 2023-08-04    12845.423973
 2023-08-05    13715.486584
 Freq: D, Name: predicted_mean, dtype: float64,
 9031.539660169708,
 7949.9743306453565)


# Create a dataframe for the metrics
sarima_metrics_df_class_monthly = pd.DataFrame({
    'RMSE': [rmse_original],
    'MAE': [mae_original]
})

print("\nSARIMA Model Monthly of Evaluation (Classic Bike)\n")
sarima_metrics_df_class_monthly

SARIMA Model Monthly of Evaluation (Classic Bike)


plt.figure(figsize=(12, 6))
plt.plot(actual_values.index, actual_values, label='Actual', color='blue')
plt.plot(original_scale_forecast.index, original_scale_forecast, label='Forecast', color='red', linestyle='--')
plt.title('Comparison of Actual and Forecasted Classic Bike Rentals with Monthly SARIMA')
plt.xlabel('Date')
plt.ylabel('Number of Rentals')
plt.legend()
plt.grid(True)
plt.show()


original_scale_forecast

2023-08-01    12440.132406
2023-08-02    12147.910183
2023-08-03    12921.383570
2023-08-04    12845.423973
2023-08-05    13715.486584
                  ...     
2024-03-26    15206.839224
2024-03-27    15193.463426
2024-03-28    15219.850350
2024-03-29    15216.632967
2024-03-30    15245.764490
Freq: D, Name: predicted_mean, Length: 243, dtype: float64


residuals = class_test_df['classic_bike_rentals'] - original_scale_forecast
residual_std = np.std(residuals)
# Calculate the 95% confidence intervals
confidence_level = 0.95
z_score = norm.ppf((1 + confidence_level) / 2)
lower_ci = original_scale_forecast - z_score * residual_std
upper_ci = original_scale_forecast + z_score * residual_std

# Create a DataFrame with the predicted values and confidence intervals
predicted_df_class_SARIMA = pd.DataFrame({'classic_bike_rentals_predict_value': original_scale_forecast,
                             'Lower_CI': lower_ci,
                             'Upper_CI': upper_ci},
                            index=class_test_df.index)


predicted_df_class_SARIMA


predicted_df_class_SARIMA.to_csv('/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/SARIMA_Classic_Prediction.csv', index = True)


plt.figure(figsize=(12, 6))
plt.plot(class_test_df.index, class_test_df['classic_bike_rentals'], label='Actual', marker='o', markersize=4)
plt.plot(predicted_df_class_SARIMA.index, predicted_df_class_SARIMA['classic_bike_rentals_predict_value'], label='Predicted', color = "red")
plt.fill_between(predicted_df_class_SARIMA.index, predicted_df_class_SARIMA['Lower_CI'],
                 predicted_df_class_SARIMA['Upper_CI'], alpha=0.2, label='95% CI', color = "gray")
plt.xlabel('Date')
plt.ylabel('Classic Bike Rentals')
plt.title('Classic Bike Predictions with 95% Confidence Interval (SARIMA Model)')
plt.legend()
plt.show()


# Plot the predictions, confidence intervals, actual values, and historical data
plt.figure(figsize=(12, 6))
plt.plot(class_df.index, class_df['classic_bike_rentals'], label='Historical Data', marker='.', markersize=4)
plt.plot(class_test_df.index, class_test_df['classic_bike_rentals'], label='Actual', marker='o', markersize=4)
plt.plot(predicted_df_class_SARIMA.index, predicted_df_class_SARIMA['classic_bike_rentals_predict_value'], label='Predicted')
plt.fill_between(predicted_df_class_SARIMA.index, predicted_df_class_SARIMA['Lower_CI'],
                 predicted_df_class_SARIMA['Upper_CI'], alpha=0.2, label='95% CI')
plt.xlabel('Date')
plt.ylabel('Classic Bike Rentals')
plt.title('Classic Bike Prediction (SARIMA Model)')
plt.legend()
plt.show()


elect_train_df=elect_train_df.dropna()


elect_train_df


# identify potential parameters by trying ACF and PACF
fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(12, 8))

# Plot the Autocorrelation Function (ACF)
plot_acf(elect_train_df['Differenced'].dropna(), lags=40, ax=ax1)

# Plot the Partial Autocorrelation Function (PACF)
plot_pacf(elect_train_df['Differenced'].dropna(), lags=40, ax=ax2)

plt.show()


model = auto_arima(elect_train_df[elect_train_df.columns[-1]], start_p=1, start_q=1,
                   test='adf',       # use adftest to find optimal 'd'
                   max_p=3, max_q=3, # maximum p and q
                   m=1,              # frequency of series
                   d=0,           # we already difference the data
                   seasonal=False,   # No Seasonality
                   start_P=0,
                   D=0,
                   trace=True,
                   error_action='ignore',
                   suppress_warnings=True,
                   stepwise=True)

print(model.summary())

Performing stepwise search to minimize aic
 ARIMA(1,0,1)(0,0,0)[0]             : AIC=18829.086, Time=0.28 sec
 ARIMA(0,0,0)(0,0,0)[0]             : AIC=19086.253, Time=0.03 sec
 ARIMA(1,0,0)(0,0,0)[0]             : AIC=19008.601, Time=0.05 sec
 ARIMA(0,0,1)(0,0,0)[0]             : AIC=18903.274, Time=0.17 sec
 ARIMA(2,0,1)(0,0,0)[0]             : AIC=18829.285, Time=0.32 sec
 ARIMA(1,0,2)(0,0,0)[0]             : AIC=18827.335, Time=0.45 sec
 ARIMA(0,0,2)(0,0,0)[0]             : AIC=18828.968, Time=0.20 sec
 ARIMA(2,0,2)(0,0,0)[0]             : AIC=18830.138, Time=0.98 sec
 ARIMA(1,0,3)(0,0,0)[0]             : AIC=18826.976, Time=0.84 sec
 ARIMA(0,0,3)(0,0,0)[0]             : AIC=18826.094, Time=0.27 sec
 ARIMA(0,0,3)(0,0,0)[0] intercept   : AIC=18827.325, Time=1.39 sec

Best model:  ARIMA(0,0,3)(0,0,0)[0]          
Total fit time: 4.984 seconds
                               SARIMAX Results                                
==============================================================================
Dep. Variable:                      y   No. Observations:                 1072
Model:               SARIMAX(0, 0, 3)   Log Likelihood               -9409.047
Date:                Mon, 13 May 2024   AIC                          18826.094
Time:                        22:25:14   BIC                          18846.003
Sample:                             0   HQIC                         18833.635
                               - 1072                                         
Covariance Type:                  opg                                         
==============================================================================
                 coef    std err          z      P>|z|      [0.025      0.975]
------------------------------------------------------------------------------
ma.L1         -0.4789      0.023    -20.921      0.000      -0.524      -0.434
ma.L2         -0.2117      0.024     -8.935      0.000      -0.258      -0.165
ma.L3         -0.0723      0.024     -2.961      0.003      -0.120      -0.024
sigma2      2.537e+06   6.19e+04     40.964      0.000    2.42e+06    2.66e+06
===================================================================================
Ljung-Box (L1) (Q):                   0.02   Jarque-Bera (JB):              1973.03
Prob(Q):                              0.90   Prob(JB):                         0.00
Heteroskedasticity (H):               4.33   Skew:                            -0.91
Prob(H) (two-sided):                  0.00   Kurtosis:                         9.39
===================================================================================

Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).


# add seasonality
s = 7  # Weekly seasonality
model_sarima_week = SARIMAX(elect_train_df['Differenced'],
                       order=(1, 0, 1),
                       seasonal_order=(1, 0, 1, s))

fitted_model_sarima_week = model_sarima_week.fit()

# Summary of the fitted SARIMA model
fitted_model_sarima_week.summary()

/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/base/tsa_model.py:473: ValueWarning: A date index has been provided, but it has no associated frequency information and so will be ignored when e.g. forecasting.
  self._init_dates(dates, freq)
/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/base/tsa_model.py:473: ValueWarning: A date index has been provided, but it has no associated frequency information and so will be ignored when e.g. forecasting.
  self._init_dates(dates, freq)


residuals = fitted_model_sarima_week.resid

fig, ax = plt.subplots(2, 2, figsize=(14, 8))

# Time series plot of residuals
ax[0, 0].plot(residuals, color='blue')
ax[0, 0].set_title('Residuals')
ax[0, 0].set_xlabel('Date')
ax[0, 0].set_ylabel('Residual')

# Histogram of residuals
ax[0, 1].hist(residuals, bins=30, color='blue', alpha=0.7)
ax[0, 1].set_title('Histogram of Residuals of Weekly SARIMA')

# Q-Q plot of residuals to check normality
sm.qqplot(residuals, line='s', ax=ax[1, 0])
ax[1, 0].set_title('Q-Q Plot of Residuals of Weekly SARIMA')

# ACF plot of residuals to check autocorrelations
sm.graphics.tsa.plot_acf(residuals, lags=40, ax=ax[1, 1])
ax[1, 1].set_title('Autocorrelation Function of Weekly SARIMA')

plt.tight_layout()
plt.show()


# Create a date range for the forecast period
forecast_start_date = elect_test_df.index.min()
forecast_end_date = elect_test_df.index.max()
forecast_dates = pd.date_range(start=forecast_start_date, end=forecast_end_date)

# Forecast
forecast = fitted_model_sarima_week.get_forecast(steps=len(forecast_dates))
forecast_mean = forecast.predicted_mean

# Since forecast_mean is a series, set its index to the forecast_dates
forecast_mean.index = forecast_dates

# Retrieve the last log value from the training set
last_log_value = elect_train_df['log_elect_bike_rentals'].iloc[-1]

# Cumulative sum of the forecasted differenced values added to the last known log value
cumulative_forecast = last_log_value + forecast_mean.cumsum()

# Exponentiate to reverse the log transformation
original_scale_forecast = np.exp(cumulative_forecast)

# Ensure elect_test_df['log_elect_bike_rentals'] exists and is correctly populated
actual_values = np.exp(elect_test_df['log_elect_bike_rentals'])

# Calculate RMSE and MAE on the original scale
rmse_original = np.sqrt(mean_squared_error(actual_values, original_scale_forecast))
mae_original = mean_absolute_error(actual_values, original_scale_forecast)

# Display the first few forecasted values, RMSE, and MAE
original_scale_forecast.head(), rmse_original, mae_original

/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/base/tsa_model.py:836: ValueWarning: No supported index is available. Prediction results will be given with an integer index beginning at `start`.
  return get_prediction_index(
/usr/local/lib/python3.10/dist-packages/statsmodels/tsa/base/tsa_model.py:836: FutureWarning: No supported index is available. In the next version, calling this method in a model without a supported index will result in an exception.
  return get_prediction_index(

(2023-07-03    5166.718810
 2023-07-04    6246.044169
 2023-07-05    7117.164546
 2023-07-06    7351.101903
 2023-07-07    7563.064900
 Freq: D, Name: predicted_mean, dtype: float64,
 3970.6662236146376,
 3350.1903935452237)


residuals = elect_test_df['elect_bike_rentals'] - original_scale_forecast
residual_std = np.std(residuals)
# Calculate the 95% confidence intervals
confidence_level = 0.95
z_score = norm.ppf((1 + confidence_level) / 2)
lower_ci = original_scale_forecast - z_score * residual_std
upper_ci = original_scale_forecast + z_score * residual_std

# Create a DataFrame with the predicted values and confidence intervals
predicted_df_elect_SARIMA = pd.DataFrame({'electric_bike_rentals_predict_value': original_scale_forecast,
                             'Lower_CI': lower_ci,
                             'Upper_CI': upper_ci},
                            index=elect_test_df.index)

predicted_df_elect_SARIMA.head()


predicted_df_elect_SARIMA.to_csv('/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/SARIMA_Electric_Prediction.csv', index = True)


# Create a dataframe for the metrics
sarima_metrics_df_elect = pd.DataFrame({
    'RMSE': [rmse_original],
    'MAE': [mae_original]
})

print("\nSARIMA Model of Evaluation (Electric Bike)\n")
sarima_metrics_df_elect

SARIMA Model of Evaluation (Electric Bike)


plt.figure(figsize=(12, 6))
plt.plot(actual_values.index, actual_values, label='Actual', color='blue')
plt.plot(original_scale_forecast.index, original_scale_forecast, label='Forecast', color='red', linestyle='--')
plt.title('Comparison of Actual and Forecasted Electric Bike Rentals with Weekly SARIMA')
plt.xlabel('Date')
plt.ylabel('Number of Rentals')
plt.legend()
plt.grid(True)
plt.show()


predicted_df_elect_SARIMA


plt.figure(figsize=(12, 6))
plt.plot(elect_test_df.index, elect_test_df['elect_bike_rentals'], label='Actual', marker='o', markersize=4)
plt.plot(predicted_df_elect_SARIMA.index, predicted_df_elect_SARIMA['electric_bike_rentals_predict_value'], label='Predicted', color = "red")
plt.fill_between(predicted_df_elect_SARIMA.index, predicted_df_elect_SARIMA['Lower_CI'],
                 predicted_df_elect_SARIMA['Upper_CI'], alpha=0.2, label='95% CI', color = "gray")
plt.xlabel('Date')
plt.ylabel('Electronic Bike Rentals')
plt.title('Electronic Bike Predictions with 95% Confidence Interval (SARIMA Model)')
plt.legend()
plt.show()


plt.figure(figsize=(12, 6))
plt.plot(elect_df.index, elect_df['elect_bike_rentals'], label='Historical Data', marker='.', markersize=4)
plt.plot(elect_test_df.index, elect_test_df['elect_bike_rentals'], label='Actual', marker='o', markersize=4)
plt.plot(predicted_df_elect_SARIMA.index, predicted_df_elect_SARIMA['electric_bike_rentals_predict_value'], label='Predicted')
plt.fill_between(predicted_df_elect_SARIMA.index, predicted_df_elect_SARIMA['Lower_CI'],
                 predicted_df_elect_SARIMA['Upper_CI'], alpha=0.2, label='95% CI')
plt.xlabel('Date')
plt.ylabel('Electric Bike Rentals')
plt.title('CElectric Bike Prediction (SARIMA Model)')
plt.legend()
plt.show()


class_df = class_df.reset_index()
class_df


split_point = pd.Timestamp('2023-08-01')
#elect_df['started_at'] = elect_df.index

train_df = class_df[class_df['started_at'] < split_point]
validation_df = class_df[class_df['started_at'] >= split_point]


print(f"Shape of elect_train_df: {train_df.shape}")
print(f"Shape of celect_test_df: {validation_df.shape}")

Shape of elect_train_df: (971, 5)
Shape of celect_test_df: (243, 5)


data=class_df[['started_at','classic_bike_rentals']]

# Generate Fourier series components
data, fs_cols = make_fourier_series_df(data, 365.25, order=5)

# Split the data
test_size = 243
train_df = data[:-test_size]
validation_df = data[-test_size:]


# Initialize MinMaxScaler
scaler_target = MinMaxScaler()


# Scale the target variable for both training and validation sets
train_df['scaled_classic_bike_rentals'] = scaler_target.fit_transform(train_df[['classic_bike_rentals']])
validation_df['scaled_classic_bike_rentals'] = scaler_target.transform(validation_df[['classic_bike_rentals']])

<ipython-input-167-94ac18fb13ce>:6: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  train_df['scaled_classic_bike_rentals'] = scaler_target.fit_transform(train_df[['classic_bike_rentals']])
<ipython-input-167-94ac18fb13ce>:7: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  validation_df['scaled_classic_bike_rentals'] = scaler_target.transform(validation_df[['classic_bike_rentals']])


%%time

# Define the objective function
def scaled_objective(global_cap, global_floor, damped_factor):
    model = DLT(
        response_col='scaled_classic_bike_rentals',
        date_col='started_at',
        estimator='stan-map',
        regressor_col=fs_cols,
        seasonality=365,
        seed=8888,
        global_trend_option='logistic',
        global_cap=global_cap,
        global_floor=global_floor,
        damped_factor=damped_factor,
        n_bootstrap_draws=1000,
        verbose=False
    )
    model.fit(df=train_df)
    predictions = model.predict(df=validation_df)
    predictions_in_original_scale = scaler_target.inverse_transform(predictions[['prediction']])


    rmse = mean_squared_error(validation_df['classic_bike_rentals'], predictions_in_original_scale, squared=False)
    return -rmse

# Parameter bounds
param_bounds = {
    'global_cap': (5000, 20000),
    'global_floor': (0, 50),
    'damped_factor': (0.01, 0.8)
}

# Setup Bayesian Optimization
optimizer = BayesianOptimization(f=scaled_objective, pbounds=param_bounds, random_state=1, verbose=2)

# Run the optimizer
optimizer.maximize(init_points=5, n_iter=15)

# Print the best parameters found
print(optimizer.max)

|   iter    |  target   | damped... | global... | global... |
-------------------------------------------------------------
| 1         | -2.226e+0 | 0.3394    | 1.58e+04  | 0.005719  |
| 2         | -2.845e+0 | 0.2488    | 7.201e+03 | 4.617     |
| 3         | -2.18e+04 | 0.1571    | 1.018e+04 | 19.84     |
| 4         | -2.074e+0 | 0.4357    | 1.129e+04 | 34.26     |
| 5         | -2.014e+0 | 0.1715    | 1.817e+04 | 1.369     |
| 6         | -1.877e+0 | 0.1532    | 1.795e+04 | 12.35     |
| 7         | -1.731e+0 | 0.7301    | 1.795e+04 | 14.91     |
| 8         | -2.548e+0 | 0.3965    | 1.795e+04 | 21.56     |
| 9         | -4.907e+0 | 0.1642    | 1.817e+04 | 5.567     |
| 10        | -2.825e+0 | 0.661     | 1.128e+04 | 29.78     |
| 11        | -1.926e+0 | 0.2036    | 1.794e+04 | 16.76     |
| 12        | -2.789e+0 | 0.1362    | 1.818e+04 | 0.3019    |
| 13        | -2.089e+0 | 0.1685    | 1.019e+04 | 13.7      |
| 14        | -1.524e+0 | 0.02061   | 1.019e+04 | 21.37     |
| 15        | -2.26e+04 | 0.5073    | 1.019e+04 | 28.34     |
| 16        | -5.013e+0 | 0.09881   | 1.02e+04  | 23.42     |
| 17        | -2.282e+0 | 0.6553    | 1.021e+04 | 15.03     |
| 18        | -2.468e+0 | 0.1751    | 1.022e+04 | 19.48     |
| 19        | -7.214e+0 | 0.8       | 1.818e+04 | 7.834     |
| 20        | -2.73e+04 | 0.5795    | 1.02e+04  | 30.39     |
=============================================================
{'target': -2789.0292110316964, 'params': {'damped_factor': 0.13620265239288523, 'global_cap': 18178.158551829154, 'global_floor': 0.3018856249946189}}
CPU times: user 20 s, sys: 3.9 s, total: 23.9 s
Wall time: 21min 55s


optimizer.max['params']

{'damped_factor': 0.13620265239288523,
 'global_cap': 18178.158551829154,
 'global_floor': 0.3018856249946189}


%%time

# Initialize and fit the model using the best parameters
model = DLT(
    response_col='scaled_classic_bike_rentals',
    date_col='started_at',
    regressor_col=fs_cols,
    estimator='stan-map',
    seasonality=365,
    seed=8888,
    global_trend_option='logistic',
    global_cap=optimizer.max['params']['global_cap'],
    global_floor=optimizer.max['params']['global_floor'],
    damped_factor=optimizer.max['params']['damped_factor'],
    n_bootstrap_draws=1000,
    verbose=False
)

model.fit(df=train_df)

# Make predictions on the validation set
# outcomes data frame
predictions = model.predict(df=validation_df)

# Inverse transform the predictions to get them back to the original scale
predictions['predictions_in_original_scale'] = scaler_target.inverse_transform(predictions[['prediction']])


# Display the predictions in the original scale
predictions


plot_predicted_data(
  training_actual_df=train_df, predicted_df=predictions,
  date_col=model.date_col, actual_col=model.response_col,
  test_actual_df=validation_df
)

WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial

<Axes: >


%%time

# Initialize and fit the model using the best parameters
model = DLT(
    response_col='scaled_classic_bike_rentals',
    date_col='started_at',
    regressor_col=fs_cols,
    estimator='stan-map',
    seasonality=365,
    seed=8888,
    global_trend_option='logistic',
    global_cap=optimizer.max['params']['global_cap'],
    global_floor=optimizer.max['params']['global_floor'],
    damped_factor=optimizer.max['params']['damped_factor'],
    n_bootstrap_draws=1000,
)

model.fit(df=train_df)

pred_df = model.predict(data)

plot_predicted_data(training_actual_df=train_df, predicted_df=pred_df,
  date_col=model.date_col, actual_col=model.response_col, test_actual_df=validation_df, lw=0.5);

2024-05-11 04:18:37 - orbit - INFO - Optimizing (CmdStanPy) with algorithm: LBFGS.
INFO:orbit:Optimizing (CmdStanPy) with algorithm: LBFGS.
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial


# If you prefer a left join to keep all rows from elect_test_df and add predictions where available
predicted_df_class_orbit = validation_df.merge(predictions[['prediction','started_at', 'predictions_in_original_scale']],
                                on='started_at', how='left')
predicted_df_class_orbit


actual = predicted_df_class_orbit['classic_bike_rentals']
predicted = predicted_df_class_orbit['predictions_in_original_scale']

# Calculating RMSE
rmse = sqrt(mean_squared_error(actual, predicted))

# Calculating MAE
mae = mean_absolute_error(actual, predicted)

# Calculating MPE
mpe = np.mean((actual - predicted) / actual) * 100

# Calculating MAPE
mape = np.mean(np.abs((actual - predicted) / actual)) * 100

print("RMSE:", rmse)
print("MAE:", mae)
print("MPE:", mpe)
print("MAPE:", mape)

RMSE: 2789.0292110316964
MAE: 2247.962273477329
MPE: 25.454942608801858
MAPE: 42.80424705456897


# Plotting actual vs. predicted values
plt.figure(figsize=(10, 5))
plt.plot(predicted_df_class_orbit['started_at'], predicted_df_class_orbit['classic_bike_rentals'] , label='Actual')
plt.plot(predicted_df_class_orbit['started_at'], predicted_df_class_orbit['predictions_in_original_scale']
, label='Predicted', linestyle='--')
plt.title('Actual vs Predicted Bike Rentals (Orbit Model)')
plt.xlabel('Date')
plt.ylabel('Number of Bike Rentals')
plt.legend()
plt.show()


# Sample data for demonstration
actuals = predicted_df_class_orbit['classic_bike_rentals']  # Replace 'actual_column_name' with the name of your actual data column
#predictions = dlt_refined.predict(df=test_df)  # This assumes predictions are already made

# Calculate residuals
predicted_df_class_orbit['residuals'] = actuals - predicted_df_class_orbit['predictions_in_original_scale']

# Manually compute CI if not available directly from model outputs
# Calculate the standard deviation of the residuals
std_residuals = np.std(predicted_df_class_orbit['residuals'])

# Calculate 95% confidence interval using the standard error
z_value = stats.norm.ppf(0.975)  # Z-score for 95% confidence

# Margin of error based on the standard deviation of residuals
margin_error = z_value * std_residuals

# Calculate lower and upper bounds of the 95% confidence interval
predicted_df_class_orbit['ci_lower'] = predicted_df_class_orbit['predictions_in_original_scale'] - margin_error
predicted_df_class_orbit['ci_upper'] = predicted_df_class_orbit['predictions_in_original_scale'] + margin_error

# Optional: Display the DataFrame to verify results
predicted_df_class_orbit[['classic_bike_rentals','predictions_in_original_scale', 'residuals', 'ci_lower', 'ci_upper']]


predicted_df_class_orbit_1 = predicted_df_class_orbit[['started_at','classic_bike_rentals','predictions_in_original_scale', 'ci_lower', 'ci_upper']]

predicted_df_class_orbit_1.head()


predicted_df_class_orbit_1.to_csv('/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/Orbit_Classic_Prediction.csv', index = False)


plt.figure(figsize=(12, 6))
plt.plot(predicted_df_class_orbit['started_at'], predicted_df_class_orbit['classic_bike_rentals'], label='Actual', color='blue')
plt.plot(predicted_df_class_orbit['started_at'], predicted_df_class_orbit['predictions_in_original_scale'], label='Predicted', color='red')
plt.fill_between(predicted_df_class_orbit['started_at'], predicted_df_class_orbit['ci_lower'],predicted_df_class_orbit['ci_upper'] , color='gray', alpha=0.2, label='95% Confidence Interval')
plt.title('Actual vs. Predicted Classic Bike Rentals with 95% CI (Orbit Model)')
plt.xlabel('Date')
plt.ylabel('Number of Bike Rentals')
plt.legend()
plt.show()


elect_df = elect_df.reset_index()
elect_df


split_point = pd.Timestamp('2023-07-03')
#elect_df['started_at'] = elect_df.index

elect_train_df = elect_df[elect_df['started_at'] < split_point]
elect_test_df = elect_df[elect_df['started_at'] >= split_point]


print(f"Shape of elect_train_df: {elect_train_df.shape}")
print(f"Shape of celect_test_df: {elect_test_df.shape}")

Shape of elect_train_df: (1073, 6)
Shape of celect_test_df: (272, 6)


data=elect_df[['started_at','elect_bike_rentals']]

# Generate Fourier series components
data, fs_cols = make_fourier_series_df(data, 365.25, order=5)

# Split the data
test_size = 272
train_df = data[:-test_size]
validation_df = data[-test_size:]


# Initialize MinMaxScaler
scaler_target = MinMaxScaler()


# Scale the target variable for both training and validation sets
train_df['scaled_elect_bike_rentals'] = scaler_target.fit_transform(train_df[['elect_bike_rentals']])
validation_df['scaled_elect_bike_rentals'] = scaler_target.transform(validation_df[['elect_bike_rentals']])

<ipython-input-191-357c25e9efde>:6: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  train_df['scaled_elect_bike_rentals'] = scaler_target.fit_transform(train_df[['elect_bike_rentals']])
<ipython-input-191-357c25e9efde>:7: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  validation_df['scaled_elect_bike_rentals'] = scaler_target.transform(validation_df[['elect_bike_rentals']])


%%time
# Define the objective function
def scaled_objective(global_cap, global_floor, damped_factor):
    model = DLT(
        response_col='scaled_elect_bike_rentals',
        date_col='started_at',
        estimator='stan-map',
        regressor_col=fs_cols,
        seasonality=365,
        seed=8888,
        global_trend_option='logistic',
        global_cap=global_cap,
        global_floor=global_floor,
        damped_factor=damped_factor,
        n_bootstrap_draws=1000,
        verbose=False
    )
    model.fit(df=train_df)
    predictions = model.predict(df=validation_df)
    predictions_in_original_scale = scaler_target.inverse_transform(predictions[['prediction']])


    rmse = mean_squared_error(validation_df['elect_bike_rentals'], predictions_in_original_scale, squared=False)
    return -rmse

# Parameter bounds
param_bounds = {
    'global_cap': (5000, 20000),
    'global_floor': (0, 50),
    'damped_factor': (0.01, 0.8)
}

# Setup Bayesian Optimization
optimizer = BayesianOptimization(f=scaled_objective, pbounds=param_bounds, random_state=1, verbose=2)

# Run the optimizer
optimizer.maximize(init_points=5, n_iter=15)

# Print the best parameters found
print(optimizer.max)

|   iter    |  target   | damped... | global... | global... |
-------------------------------------------------------------

/usr/local/lib/python3.10/dist-packages/orbit/forecaster/forecaster.py:290: UserWarning: Datetime index is not evenly distributed
  warnings.warn("Datetime index is not evenly distributed")

| 1         | -1.387e+0 | 0.3394    | 1.58e+04  | 0.005719  |

/usr/local/lib/python3.10/dist-packages/orbit/forecaster/forecaster.py:290: UserWarning: Datetime index is not evenly distributed
  warnings.warn("Datetime index is not evenly distributed")

| 2         | -8.74e+03 | 0.2488    | 7.201e+03 | 4.617     |

/usr/local/lib/python3.10/dist-packages/orbit/forecaster/forecaster.py:290: UserWarning: Datetime index is not evenly distributed
  warnings.warn("Datetime index is not evenly distributed")

| 3         | -7.646e+0 | 0.1571    | 1.018e+04 | 19.84     |

/usr/local/lib/python3.10/dist-packages/orbit/forecaster/forecaster.py:290: UserWarning: Datetime index is not evenly distributed
  warnings.warn("Datetime index is not evenly distributed")

| 4         | -9.716e+0 | 0.4357    | 1.129e+04 | 34.26     |

/usr/local/lib/python3.10/dist-packages/orbit/forecaster/forecaster.py:290: UserWarning: Datetime index is not evenly distributed
  warnings.warn("Datetime index is not evenly distributed")


optimizer.max['params']

{'damped_factor': 0.1588582927313985,
 'global_cap': 6993.568718733896,
 'global_floor': 8.86597973211151}


%%time

# Initialize and fit the model using the best parameters
model = DLT(
    response_col='scaled_elect_bike_rentals',
    date_col='started_at',
    regressor_col=fs_cols,
    estimator='stan-map',
    seasonality=365,
    seed=8888,
    global_trend_option='logistic',
    global_cap=optimizer.max['params']['global_cap'],
    global_floor=optimizer.max['params']['global_floor'],
    damped_factor=optimizer.max['params']['damped_factor'],
    n_bootstrap_draws=1000,
    verbose=False
)

model.fit(df=train_df)

# Make predictions on the validation set
# outcomes data frame
predictions = model.predict(df=validation_df)

# Inverse transform the predictions to get them back to the original scale
predictions['predictions_in_original_scale'] = scaler_target.inverse_transform(predictions[['prediction']])


# Display the predictions in the original scale
predictions

/usr/local/lib/python3.10/dist-packages/orbit/forecaster/forecaster.py:290: UserWarning: Datetime index is not evenly distributed
  warnings.warn("Datetime index is not evenly distributed")

CPU times: user 896 ms, sys: 10 ms, total: 906 ms
Wall time: 4.9 s


plot_predicted_data(
  training_actual_df=train_df, predicted_df=predictions,
  date_col=model.date_col, actual_col=model.response_col,
  test_actual_df=validation_df
)

WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial

<Axes: >


# Initialize and fit the model using the best parameters
model = DLT(
    response_col='scaled_elect_bike_rentals',
    date_col='started_at',
    regressor_col=fs_cols,
    estimator='stan-map',
    seasonality=365,
    seed=8888,
    global_trend_option='logistic',
    global_cap=optimizer.max['params']['global_cap'],
    global_floor=optimizer.max['params']['global_floor'],
    damped_factor=optimizer.max['params']['damped_factor'],
    n_bootstrap_draws=1000,
    verbose=False
)

model.fit(df=train_df)

pred_df = model.predict(data)

plot_predicted_data(training_actual_df=train_df, predicted_df=pred_df,
  date_col=model.date_col, actual_col=model.response_col, test_actual_df=validation_df, lw=0.5);

/usr/local/lib/python3.10/dist-packages/orbit/forecaster/forecaster.py:290: UserWarning: Datetime index is not evenly distributed
  warnings.warn("Datetime index is not evenly distributed")
/usr/local/lib/python3.10/dist-packages/orbit/forecaster/forecaster.py:335: UserWarning: Datetime index is not evenly distributed
  warnings.warn("Datetime index is not evenly distributed")
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial
WARNING:matplotlib.font_manager:findfont: Generic family 'sans-serif' not found because none of the following families were found: Arial


# If you prefer a left join to keep all rows from elect_test_df and add predictions where available
predicted_df_elect_orbit = validation_df.merge(predictions[['prediction','started_at', 'predictions_in_original_scale']],
                                on='started_at', how='left')
predicted_df_elect_orbit


actual = predicted_df_elect_orbit['elect_bike_rentals']
predicted = predicted_df_elect_orbit['predictions_in_original_scale']

# Calculating RMSE
rmse = sqrt(mean_squared_error(actual, predicted))

# Calculating MAE
mae = mean_absolute_error(actual, predicted)

# Calculating MPE
mpe = np.mean((actual - predicted) / actual) * 100

# Calculating MAPE
mape = np.mean(np.abs((actual - predicted) / actual)) * 100

print("RMSE:", rmse)
print("MAE:", mae)
print("MPE:", mpe)
print("MAPE:", mape)

RMSE: 6428.848968091197
MAE: 4714.146791080093
MPE: -37.331486984296454
MAPE: 151.78190001413603


# Plotting actual vs. predicted values
plt.figure(figsize=(10, 5))
plt.plot(predicted_df_elect_orbit['started_at'], predicted_df_elect_orbit['elect_bike_rentals'] , label='Actual')
plt.plot(predicted_df_elect_orbit['started_at'], predicted_df_elect_orbit['predictions_in_original_scale']
, label='Predicted', linestyle='--')
plt.title('Actual vs Predicted Bike Rentals (Orbit Model)')
plt.xlabel('Date')
plt.ylabel('Number of Bike Rentals')
plt.legend()
plt.show()


# Sample data for demonstration
actuals = predicted_df_elect_orbit['elect_bike_rentals']  # Replace 'actual_column_name' with the name of your actual data column
#predictions = dlt_refined.predict(df=test_df)  # This assumes predictions are already made

# Calculate residuals
predicted_df_elect_orbit['residuals'] = actuals - predicted_df_elect_orbit['predictions_in_original_scale']

# Manually compute CI if not available directly from model outputs
# Calculate the standard deviation of the residuals
std_residuals = np.std(predicted_df_elect_orbit['residuals'])

# Calculate 95% confidence interval using the standard error
z_value = stats.norm.ppf(0.975)  # Z-score for 95% confidence

# Margin of error based on the standard deviation of residuals
margin_error = z_value * std_residuals

# Calculate lower and upper bounds of the 95% confidence interval
predicted_df_elect_orbit['ci_lower'] = predicted_df_elect_orbit['predictions_in_original_scale'] - margin_error
predicted_df_elect_orbit['ci_upper'] = predicted_df_elect_orbit['predictions_in_original_scale'] + margin_error

# Optional: Display the DataFrame to verify results
predicted_df_elect_orbit[['elect_bike_rentals','predictions_in_original_scale', 'residuals', 'ci_lower', 'ci_upper']]


predicted_df_elect_orbit_1 = predicted_df_elect_orbit[['started_at','elect_bike_rentals','predictions_in_original_scale', 'ci_lower', 'ci_upper']]
predicted_df_elect_orbit_1.head()


predicted_df_elect_orbit_1.to_csv('/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/Orbit_Electric_Prediction.csv', index = False)


plt.figure(figsize=(12, 6))
plt.plot(predicted_df_elect_orbit['started_at'], predicted_df_elect_orbit['elect_bike_rentals'], label='Actual', color='blue')
plt.plot(predicted_df_elect_orbit['started_at'], predicted_df_elect_orbit['predictions_in_original_scale'], label='Predicted', color='red')
plt.fill_between(predicted_df_elect_orbit['started_at'], predicted_df_elect_orbit['ci_lower'],predicted_df_elect_orbit['ci_upper'] , color='gray', alpha=0.2, label='95% Confidence Interval')
plt.title('Actual vs. Predicted Bike Rentals with 95% CI (Obrit Model)')
plt.xlabel('Date')
plt.ylabel('Number of Bike Rentals')
plt.legend()
plt.show()


df_reset = class_train_df.reset_index()
df_prophet = df_reset.rename(columns={'started_at': 'ds', 'classic_bike_rentals': 'y'})

# Instantiate and fit the model with Prophet
model = Prophet(daily_seasonality=True, weekly_seasonality=False, yearly_seasonality=True)
model.add_seasonality(name='weekly', period=7, fourier_order=3)
model = Prophet(daily_seasonality=True)
model.fit(df_prophet)

DEBUG:cmdstanpy:input tempfile: /tmp/tmpibf8dki5/mp0vxa6p.json
DEBUG:cmdstanpy:input tempfile: /tmp/tmpibf8dki5/ojhd96mj.json
DEBUG:cmdstanpy:idx 0
DEBUG:cmdstanpy:running CmdStan, num_threads: None
DEBUG:cmdstanpy:CmdStan args: ['/usr/local/lib/python3.10/dist-packages/prophet/stan_model/prophet_model.bin', 'random', 'seed=31891', 'data', 'file=/tmp/tmpibf8dki5/mp0vxa6p.json', 'init=/tmp/tmpibf8dki5/ojhd96mj.json', 'output', 'file=/tmp/tmpibf8dki5/prophet_model280hwhno/prophet_model-20240513044929.csv', 'method=optimize', 'algorithm=lbfgs', 'iter=10000']
04:49:29 - cmdstanpy - INFO - Chain [1] start processing
INFO:cmdstanpy:Chain [1] start processing
04:49:29 - cmdstanpy - INFO - Chain [1] done processing
INFO:cmdstanpy:Chain [1] done processing

<prophet.forecaster.Prophet at 0x7918e5b785e0>


# Extend the future DataFrame with necessary regressors
future = model.make_future_dataframe(periods=len(class_test_df), freq='D')

# Assuming last available y value is the last observation in the training set
last_known_y = df_prophet['y'].iloc[-1]

# We simulate future 'y' values by carrying forward the last known value
# This is a placeholder; you might want to use a better method in practice
future['y_lag1'] = last_known_y

# Predicting the future with the model
forecast = model.predict(future)

# Correct calculation for confidence intervals using 'yhat' and the z-score
confidence_level = 0.95
z_score = norm.ppf((1 + confidence_level) / 2)
forecast['Lower_CI'] = forecast['yhat'] - z_score * forecast['yhat_lower']
forecast['Upper_CI'] = forecast['yhat'] + z_score * forecast['yhat_upper']

# Save the results into a new DataFrame
results_df_prophet = forecast[['ds', 'yhat', 'Lower_CI', 'Upper_CI']].rename(columns={'ds': 'started_at', 'yhat': 'predicted_classic_bike_rentals_prophet'})

# Plotting
fig1 = model.plot(forecast)
test_data_for_plotting = class_test_df.copy()
test_data_for_plotting['started_at'] = pd.to_datetime(test_data_for_plotting.index)

plt.scatter(test_data_for_plotting['started_at'], test_data_for_plotting['classic_bike_rentals'], color='red', marker='.', label='Test Data')
last_train_date = df_prophet['ds'].max()
plt.axvline(x=last_train_date, color='gray', linestyle='--', linewidth=2, label='End of Training Data')
plt.legend()
plt.show()


results_df_prophet.to_csv('/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/Prophet Classic.csv')


from sklearn.metrics import mean_squared_error, mean_absolute_error, mean_absolute_percentage_error
forecast['ds'] = pd.to_datetime(forecast['ds'])

# Align the forecasted data with the test dataset dates
forecast_test_period = forecast.set_index('ds')['yhat'].loc[class_test_df.index]

# Calculate the errors
mae = mean_absolute_error(class_test_df['classic_bike_rentals'], forecast_test_period)
rmse = np.sqrt(mean_squared_error(class_test_df['classic_bike_rentals'], forecast_test_period))
print(f"Mean Absolute Error (MAE): {mae}")
print(f"Root Mean Square Error (RMSE): {rmse}")

Mean Absolute Error (MAE): 2517.2092464621983
Root Mean Square Error (RMSE): 3030.3381626993587


residuals_class = class_test_df['classic_bike_rentals'] - forecast_test_period
# Plot residuals
plt.figure(figsize=(10, 5))
plt.plot(residuals_class.index, residuals_class, label='Residuals')
plt.axhline(y=0, color='r', linestyle='--')
plt.title('Residuals of Forecast Classic')
plt.xlabel('Date')
plt.ylabel('Residuals')
plt.legend()
plt.show()


plot_acf(residuals_class, lags=30, alpha=0.05)
plt.title('ACF of Residuals Classic')
plt.show()


plot_pacf(residuals_class, lags=30, alpha=0.05)
plt.title('PACF of Residuals Classic')
plt.show()


df_reset_elect = elect_train_df.reset_index()
df_prophet_elect = df_reset_elect.rename(columns={'started_at': 'ds', 'elect_bike_rentals': 'y'})
model_elect = Prophet(daily_seasonality=True, weekly_seasonality=False, yearly_seasonality=True)
model_elect.add_seasonality(name='weekly', period=7, fourier_order=3)
model_elect.fit(df_prophet_elect)

DEBUG:cmdstanpy:input tempfile: /tmp/tmpibf8dki5/kiynq6wb.json
DEBUG:cmdstanpy:input tempfile: /tmp/tmpibf8dki5/f1yd8hm5.json
DEBUG:cmdstanpy:idx 0
DEBUG:cmdstanpy:running CmdStan, num_threads: None
DEBUG:cmdstanpy:CmdStan args: ['/usr/local/lib/python3.10/dist-packages/prophet/stan_model/prophet_model.bin', 'random', 'seed=28619', 'data', 'file=/tmp/tmpibf8dki5/kiynq6wb.json', 'init=/tmp/tmpibf8dki5/f1yd8hm5.json', 'output', 'file=/tmp/tmpibf8dki5/prophet_model9bqlygx2/prophet_model-20240513044959.csv', 'method=optimize', 'algorithm=lbfgs', 'iter=10000']
04:49:59 - cmdstanpy - INFO - Chain [1] start processing
INFO:cmdstanpy:Chain [1] start processing
04:49:59 - cmdstanpy - INFO - Chain [1] done processing
INFO:cmdstanpy:Chain [1] done processing

<prophet.forecaster.Prophet at 0x79179cbc1bd0>


num_periods_to_predict = len(elect_test_df)
future_elect = model_elect.make_future_dataframe(periods=num_periods_to_predict, freq='D')

forecast_elect = model_elect.predict(future_elect)
num_periods_to_predict = len(elect_test_df)
future_elect = model_elect.make_future_dataframe(periods=num_periods_to_predict, freq='D')

forecast_elect = model_elect.predict(future_elect)

# Calculate the 95% confidence intervals
confidence_level = 0.95
z_score = norm.ppf((1 + confidence_level) / 2)
forecast_elect['Lower_CI'] = forecast_elect['yhat'] - z_score * forecast_elect['yhat_lower']
forecast_elect['Upper_CI'] = forecast_elect['yhat'] + z_score * forecast_elect['yhat_upper']

# Save the results into a new DataFrame
results_df_prophet_elect = forecast_elect[['ds', 'yhat', 'Lower_CI', 'Upper_CI']].rename(columns={'ds': 'started_at', 'yhat': 'predicted_elect_bike_rentals_prophet'})


results_df_prophet_elect.to_csv('/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/Prophet Elect Results.csv')


fig2 = model_elect.plot(forecast_elect)

test_data_for_plotting_elect = elect_test_df.reset_index()

plt.scatter(test_data_for_plotting_elect['started_at'], test_data_for_plotting_elect['elect_bike_rentals'], color='orange', marker='.', label='Test Data')

last_train_date_elect = df_prophet_elect['ds'].max()
plt.axvline(x=last_train_date_elect, color='gray', linestyle='--', linewidth=2, label='End of Training Data')

plt.legend()
plt.show()


forecast_elect['ds'] = pd.to_datetime(forecast_elect['ds'])

# Align the forecasted data with the test dataset dates
forecast_test_period_elect = forecast_elect.set_index('ds')['yhat'].loc[elect_test_df.index]

# Calculate the errors using the 'Differenced' data column from the test set
mae_elect = mean_absolute_error(elect_test_df['elect_bike_rentals'], forecast_test_period_elect)
rmse_elect = np.sqrt(mean_squared_error(elect_test_df['elect_bike_rentals'], forecast_test_period_elect))

# MAE and RMSE values
print(f"Mean Absolute Error (MAE) on Elect Data: {mae_elect}")
print(f"Root Mean Square Error (RMSE) on Elect Data: {rmse_elect}")

Mean Absolute Error (MAE) on Elect Data: 2931.6149396636956
Root Mean Square Error (RMSE) on Elect Data: 3415.465331171118


residuals_elect = elect_test_df['elect_bike_rentals'] - forecast_test_period_elect

plt.figure(figsize=(10, 5))
plt.plot(residuals_elect.index, residuals_elect, label='Residuals')
plt.axhline(y=0, color='r', linestyle='--')
plt.title('Residuals of Forecast Electronic')
plt.xlabel('Date')
plt.ylabel('Residuals')
plt.legend()
plt.show()


# ACF plot of residuals
plot_acf(residuals_elect, lags=30, alpha=0.05)
plt.title('ACF of Residuals Electronic')
plt.show()


# PACF plot of residuals
plot_pacf(residuals_elect, lags=30, alpha=0.05)
plt.title('PACF of Residuals Electronic')
plt.show()


sarima_class_df = pd.read_csv("/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/SARIMA_Classic_Prediction.csv", index_col = False)
sarima_elect_df = pd.read_csv("/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/SARIMA_Electric_Prediction.csv", index_col = False)
lstm_class_df = pd.read_csv("/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/predicted_df_classic_lstms.csv", index_col = False)
lstm_elect_df = pd.read_csv("/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/predicted_df_electric_lstms.csv")
lstm_elect_df = lstm_elect_df.drop(columns=['Unnamed: 0'])
orbit_class_df = pd.read_csv("/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/Orbit_Classic_Prediction.csv")
orbit_elect_df = pd.read_csv("/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/Orbit_Electric_Prediction.csv")
fb_class_df = pd.read_csv("/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/Prophet Classic.csv")
fb_class_df = fb_class_df.drop(columns=['Unnamed: 0'])
fb_elect_df = pd.read_csv("/content/drive/Shareddrives/time series/(Team) ADSP 31006 Time Series Analysis and Forecasting/Prediction Result/Prophet Elect Results.csv")
fb_elect_df = fb_elect_df.drop(columns=['Unnamed: 0'])


lstm_class_df.rename(columns={'Start_Test_Date': 'started_at'}, inplace=True)

merged_df_class = sarima_class_df.merge(lstm_class_df, on='started_at', how='inner')
merged_df_class = merged_df_class.merge(orbit_class_df, on='started_at', how='inner')
merged_df_class = merged_df_class.merge(fb_class_df, on='started_at', how='inner')

plt.style.use('seaborn-darkgrid')
fig, ax = plt.subplots(figsize=(16, 8))

ax.plot(merged_df_class['started_at'], merged_df_class['classic_bike_rentals_predict_value'], label='SARIMA Predicted Value', linewidth=2)
ax.plot(merged_df_class['started_at'], merged_df_class['Classic_Predicted_Value'], label='LSTM Predicted Value', linewidth=2)
ax.plot(merged_df_class['started_at'], merged_df_class['predictions_in_original_scale'], label='Orbit Predicted Value', linewidth=2)
ax.plot(merged_df_class['started_at'], merged_df_class['predicted_classic_bike_rentals_prophet'], label='Prophet Predicted Value', linewidth=2)

ax.plot(merged_df_class['started_at'], merged_df_class['Classic_Actual_Value'], label='Classic Bike Actual Value', color='black', linestyle='--', linewidth=2)

ax.set_xlabel('Date', fontsize=12)
ax.set_ylabel('Predicted Values', fontsize=12)
ax.set_title('Comparison of Predicted Values from Different Models (Classic Bike)', fontsize=15)

ax.tick_params(axis='x', which='major', labelsize=12)
ax.tick_params(axis='y', which='major', labelsize=12)

ax.xaxis.set_major_locator(plt.MaxNLocator(nbins=6))
plt.xticks(rotation=45)
ax.legend(loc='upper left', bbox_to_anchor=(1, 0.5))
ax.grid(True)
plt.tight_layout()
plt.show()

<ipython-input-69-9117bee6b0a7>:7: MatplotlibDeprecationWarning: The seaborn styles shipped by Matplotlib are deprecated since 3.6, as they no longer correspond to the styles shipped by seaborn. However, they will remain available as 'seaborn-v0_8-<style>'. Alternatively, directly use the seaborn API instead.
  plt.style.use('seaborn-darkgrid')


lstm_elect_df.rename(columns={'Start_Test_Date': 'started_at'}, inplace=True)
merged_df_elect = sarima_elect_df.merge(lstm_elect_df, on='started_at', how='inner')
merged_df_elect = merged_df_elect.merge(orbit_elect_df, on='started_at', how='inner')
merged_df_elect = merged_df_elect.merge(fb_elect_df, on='started_at', how='inner')

plt.style.use('seaborn-darkgrid')

fig, ax = plt.subplots(figsize=(16, 8))

ax.plot(merged_df_elect['started_at'], merged_df_elect['electric_bike_rentals_predict_value'], label='SARIMA Predicted Value', linewidth=2)
ax.plot(merged_df_elect['started_at'], merged_df_elect['Electric_Predicted_Value'], label='LSTM Predicted Value', linewidth=2)
ax.plot(merged_df_elect['started_at'], merged_df_elect['predictions_in_original_scale'], label='Orbit Predicted Value', linewidth=2)
ax.plot(merged_df_elect['started_at'], merged_df_elect['predicted_elect_bike_rentals_prophet'], label='Prophet Predicted Value', linewidth=2)

ax.plot(merged_df_elect['started_at'], merged_df_elect['Electric_Actual_Value'], label='Electric Bike Actual Value', color='black', linestyle='--', linewidth=2)
ax.set_xlabel('Date', fontsize=12)
ax.set_ylabel('Predicted Values', fontsize=12)
ax.set_title('Comparison of Predicted Values from Different Models (Electric Bike)', fontsize=15)
ax.tick_params(axis='x', which='major', labelsize=12)
ax.tick_params(axis='y', which='major', labelsize=12)
ax.xaxis.set_major_locator(plt.MaxNLocator(nbins=6))
plt.xticks(rotation=45)
ax.legend(loc='upper left', bbox_to_anchor=(1, 0.5))
ax.grid(True)
plt.tight_layout()
plt.show()

<ipython-input-71-1a1516e298a3>:6: MatplotlibDeprecationWarning: The seaborn styles shipped by Matplotlib are deprecated since 3.6, as they no longer correspond to the styles shipped by seaborn. However, they will remain available as 'seaborn-v0_8-<style>'. Alternatively, directly use the seaborn API instead.
  plt.style.use('seaborn-darkgrid')

	ride_id	rideable_type	started_at	ended_at	start_station_name	start_station_id	end_station_name	end_station_id	start_lat	start_lng	end_lat	end_lng	member_casual
0	EACB19130B0CDA4A	docked_bike	2020-01-21 20:06:59	2020-01-21 20:14:30	Western Ave & Leland Ave	239	Clark St & Leland Ave	326.0	41.9665	-87.6884	41.9671	-87.6674	member
1	8FED874C809DC021	docked_bike	2020-01-30 14:22:39	2020-01-30 14:26:22	Clark St & Montrose Ave	234	Southport Ave & Irving Park Rd	318.0	41.9616	-87.6660	41.9542	-87.6644	member
2	789F3C21E472CA96	docked_bike	2020-01-09 19:29:26	2020-01-09 19:32:17	Broadway & Belmont Ave	296	Wilton Ave & Belmont Ave	117.0	41.9401	-87.6455	41.9402	-87.6530	member
3	C9A388DAC6ABF313	docked_bike	2020-01-06 16:17:07	2020-01-06 16:25:56	Clark St & Randolph St	51	Fairbanks Ct & Grand Ave	24.0	41.8846	-87.6319	41.8918	-87.6206	member
4	943BC3CBECCFD662	docked_bike	2020-01-30 08:37:16	2020-01-30 08:42:48	Clinton St & Lake St	66	Wells St & Hubbard St	212.0	41.8856	-87.6418	41.8899	-87.6343	member

	ride_id	rideable_type	started_at	ended_at	start_station_name	start_station_id	end_station_name	end_station_id	start_lat	start_lng	end_lat	end_lng	member_casual
0	A847FADBBC638E45	docked_bike	2020-04-26 17:45:14	2020-04-26 18:12:03	Eckhart Park	86	Lincoln Ave & Diversey Pkwy	152.0	41.8964	-87.6610	41.9322	-87.6586	member
1	5405B80E996FF60D	docked_bike	2020-04-17 17:08:54	2020-04-17 17:17:03	Drake Ave & Fullerton Ave	503	Kosciuszko Park	499.0	41.9244	-87.7154	41.9306	-87.7238	member
2	5DD24A79A4E006F4	docked_bike	2020-04-01 17:54:13	2020-04-01 18:08:36	McClurg Ct & Erie St	142	Indiana Ave & Roosevelt Rd	255.0	41.8945	-87.6179	41.8679	-87.6230	member
3	2A59BBDF5CDBA725	docked_bike	2020-04-07 12:50:19	2020-04-07 13:02:31	California Ave & Division St	216	Wood St & Augusta Blvd	657.0	41.9030	-87.6975	41.8992	-87.6722	member
4	27AD306C119C6158	docked_bike	2020-04-18 10:22:59	2020-04-18 11:15:54	Rush St & Hubbard St	125	Sheridan Rd & Lawrence Ave	323.0	41.8902	-87.6262	41.9695	-87.6547	casual

	ride_id	rideable_type	started_at	ended_at	start_station_name	start_station_id	end_station_name	end_station_id	start_lat	start_lng	end_lat	end_lng	member_casual
0	02668AD35674B983	docked_bike	2020-05-27 10:03:52	2020-05-27 10:16:49	Franklin St & Jackson Blvd	36	Wabash Ave & Grand Ave	199.0	41.8777	-87.6353	41.8915	-87.6268	member
1	7A50CCAF1EDDB28F	docked_bike	2020-05-25 10:47:11	2020-05-25 11:05:40	Clark St & Wrightwood Ave	340	Clark St & Leland Ave	326.0	41.9295	-87.6431	41.9671	-87.6674	casual
2	2FFCDFDB91FE9A52	docked_bike	2020-05-02 14:11:03	2020-05-02 15:48:21	Kedzie Ave & Milwaukee Ave	260	Kedzie Ave & Milwaukee Ave	260.0	41.9296	-87.7079	41.9296	-87.7079	casual
3	58991CF1DB75BA84	docked_bike	2020-05-02 16:25:36	2020-05-02 16:39:28	Clarendon Ave & Leland Ave	251	Lake Shore Dr & Wellington Ave	157.0	41.9680	-87.6500	41.9367	-87.6368	casual
4	A79651EFECC268CD	docked_bike	2020-05-29 12:49:54	2020-05-29 13:27:11	Hermitage Ave & Polk St	261	Halsted St & Archer Ave	206.0	41.8715	-87.6699	41.8472	-87.6468	member

	lb_stat	lb_pvalue
5	2.224862	0.817237
10	3.700074	0.959865
15	4.896029	0.992971
20	5.979182	0.998925

	Start_Test_Date	Classic_Actual_Value	Classic_Predicted_Value	Lower CI	Upper CI
0	2023-09-02	15042.0	13020.050781	9699.286133	16340.815430
1	2023-09-03	14795.0	13566.754883	10245.990234	16887.519531
2	2023-09-04	13181.0	12405.963867	9085.199219	15726.728516
3	2023-09-05	11716.0	11316.838867	7996.074219	14637.603516
4	2023-09-06	10307.0	10956.805664	7636.041016	14277.570312

	elect_bike_rentals	day
started_at
2020-07-13	8	Monday
2020-07-14	14	Tuesday
2020-07-15	13	Wednesday
2020-07-16	0	Thursday
2020-07-17	0	Friday
...	...	...
2024-03-26	5208	Tuesday
2024-03-27	5312	Wednesday
2024-03-28	6564	Thursday
2024-03-29	5585	Friday
2024-03-30	5273	Saturday

	lb_stat	lb_pvalue
5	0.727194	0.981443
10	0.800631	0.999939
15	0.906831	1.000000
20	1.042905	1.000000

	Start_Test_Date	Electric_Actual_Value	Electric_Predicted_Value	Lower CI	Upper CI
0	2023-08-04	14987.0	13596.238281	10636.766602	16555.710938
1	2023-08-05	8680.0	13770.071289	10810.599609	16729.542969
2	2023-08-06	9101.0	10825.515625	7866.043945	13784.987305
3	2023-08-07	10676.0	10683.356445	7723.884766	13642.828125
4	2023-08-08	11941.0	11529.010742	8569.539062	14488.482422

Dep. Variable:	Differenced	No. Observations:	971
Model:	SARIMAX(1, 0, 1)x(1, 0, 1, 7)	Log Likelihood	-396.977
Date:	Mon, 13 May 2024	AIC	803.955
Time:	22:25:04	BIC	828.346
Sample:	12-03-2020	HQIC	813.238
	- 07-31-2023
Covariance Type:	opg

	coef	std err	z	P>\|z\|	[0.025	0.975]
ar.L1	0.4582	0.035	13.122	0.000	0.390	0.527
ma.L1	-0.8663	0.022	-39.660	0.000	-0.909	-0.824
ar.S.L7	0.9015	0.054	16.647	0.000	0.795	1.008
ma.S.L7	-0.8333	0.068	-12.306	0.000	-0.966	-0.701
sigma2	0.1325	0.004	35.223	0.000	0.125	0.140

Ljung-Box (L1) (Q):	0.23	Jarque-Bera (JB):	497.85
Prob(Q):	0.63	Prob(JB):	0.00
Heteroskedasticity (H):	0.93	Skew:	-0.72
Prob(H) (two-sided):	0.51	Kurtosis:	6.20

	classic_bike_rentals_predict_value	Lower_CI	Upper_CI
started_at
2023-08-01	12440.132406	3276.022227	21604.242585
2023-08-02	12147.910183	2983.800004	21312.020362
2023-08-03	12921.383570	3757.273391	22085.493749
2023-08-04	12845.423973	3681.313794	22009.534152
2023-08-05	13715.486584	4551.376405	22879.596763
...	...	...	...
2024-03-26	15206.839224	6042.729045	24370.949403
2024-03-27	15193.463426	6029.353247	24357.573605
2024-03-28	15219.850350	6055.740171	24383.960530
2024-03-29	15216.632967	6052.522788	24380.743146
2024-03-30	15245.764490	6081.654310	24409.874669

Ljung-Box (L1) (Q):	0.05	Jarque-Bera (JB):	652162.49
Prob(Q):	0.83	Prob(JB):	0.00
Heteroskedasticity (H):	1.87	Skew:	-6.02
Prob(H) (two-sided):	0.00	Kurtosis:	123.23

	electric_bike_rentals_predict_value	Lower_CI	Upper_CI
started_at
2023-07-03	5166.718810	-2194.837327	12528.274946
2023-07-04	6246.044169	-1115.511968	13607.600305
2023-07-05	7117.164546	-244.391590	14478.720683
2023-07-06	7351.101903	-10.454233	14712.658040
2023-07-07	7563.064900	201.508764	14924.621037

	started_at	classic_bike_rentals	day	log_classic_bike_rentals	Differenced
0	2020-12-03	669	Thursday	6.505784	0.798674
1	2020-12-04	3190	Friday	8.067776	1.561992
2	2020-12-05	3586	Saturday	8.184793	0.117016
3	2020-12-06	2485	Sunday	7.818028	-0.366765
4	2020-12-07	2915	Monday	7.977625	0.159597
...	...	...	...	...	...
1209	2024-03-26	3973	Tuesday	8.287277	-0.170166
1210	2024-03-27	4451	Wednesday	8.400884	0.113607
1211	2024-03-28	5370	Thursday	8.588583	0.187699
1212	2024-03-29	4933	Friday	8.503703	-0.084881
1213	2024-03-30	5228	Saturday	8.561784	0.058081

	elect_bike_rentals	day	log_elect_bike_rentals	Differenced	diff_elect_bike_rentals
started_at
2020-07-14	14	Tuesday	2.639057	0.559616	6.0
2020-07-15	13	Wednesday	2.564949	-0.074108	-1.0
2020-07-28	31	Tuesday	3.433987	0.869038	18.0
2020-07-29	287	Wednesday	5.659482	2.225495	256.0
2020-07-30	745	Thursday	6.613384	0.953902	458.0
...	...	...	...	...	...
2023-06-28	12073	Wednesday	9.398727	0.100650	1156.0
2023-06-29	11045	Thursday	9.309733	-0.088994	-1028.0
2023-06-30	13507	Friday	9.510963	0.201230	2462.0
2023-07-01	11817	Saturday	9.377294	-0.133669	-1690.0
2023-07-02	3351	Sunday	8.117014	-1.260280	-8466.0

	coef	std err	z	P>\|z\|	[0.025	0.975]
ar.L1	0.4072	0.036	11.393	0.000	0.337	0.477
ma.L1	-0.7536	0.033	-22.665	0.000	-0.819	-0.688
ar.S.L7	0.8881	0.062	14.425	0.000	0.767	1.009
ma.S.L7	-0.8045	0.068	-11.781	0.000	-0.938	-0.671
sigma2	0.1626	0.001	163.482	0.000	0.161	0.165

	started_at	prediction_5	prediction	prediction_95	predictions_in_original_scale
0	2023-08-01	0.175782	0.475162	0.787967	12359.100699
1	2023-08-02	0.150073	0.449196	0.756565	11695.348159
2	2023-08-03	0.059904	0.379138	0.697189	9904.524350
3	2023-08-04	0.161112	0.454362	0.787037	11827.405121
4	2023-08-05	0.171512	0.500230	0.777777	12999.873125
...	...	...	...	...	...
238	2024-03-26	-0.289310	0.102959	0.509105	2844.847180
239	2024-03-27	-0.299329	0.126157	0.529326	3437.817481
240	2024-03-28	-0.315916	0.105382	0.505502	2906.787308
241	2024-03-29	-0.284755	0.111501	0.529302	3063.189961
242	2024-03-30	-0.348671	0.078053	0.470930	2208.191865

	started_at	classic_bike_rentals	fs_cos1	fs_sin1	fs_cos2	fs_sin2	fs_cos3	fs_sin3	fs_cos4	fs_sin4	fs_cos5	fs_sin5	scaled_classic_bike_rentals	prediction	predictions_in_original_scale
0	2023-08-01	13573	-0.529494	-0.848314	-0.439273	0.898354	0.994678	-1.030314e-01	-0.614079	-0.789245	-0.344376	0.938832	0.522651	0.475162	12359.100699
1	2023-08-02	14901	-0.514823	-0.857296	-0.469915	0.882712	0.998669	-5.158437e-02	-0.558361	-0.829598	-0.423755	0.905777	0.574603	0.449196	11695.348159
2	2023-08-03	14974	-0.500000	-0.866025	-0.500000	0.866025	1.000000	-1.959435e-15	-0.500000	-0.866025	-0.500000	0.866025	0.577459	0.379138	9904.524350
3	2023-08-04	16130	-0.485029	-0.874498	-0.529494	0.848314	0.998669	5.158437e-02	-0.439273	-0.898354	-0.572548	0.819871	0.622682	0.454362	11827.405121
4	2023-08-05	8218	-0.469915	-0.882712	-0.558361	0.829598	0.994678	1.030314e-01	-0.376467	-0.926430	-0.640864	0.767655	0.313160	0.500230	12999.873125
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
238	2024-03-26	3973	-0.384421	0.923158	-0.704441	-0.709763	0.926025	-3.774625e-01	-0.007526	0.999972	-0.920239	-0.391358	0.147093	0.102959	2844.847180
239	2024-03-27	4451	-0.400244	0.916409	-0.679610	-0.733574	0.944263	-3.291915e-01	-0.076262	0.997088	-0.883217	-0.468965	0.165793	0.126157	3437.817481
240	2024-03-28	5370	-0.415948	0.909388	-0.653974	-0.756517	0.959987	-2.800440e-01	-0.144636	0.989485	-0.839665	-0.543105	0.201745	0.105382	2906.787308
241	2024-03-29	4933	-0.431530	0.902099	-0.627564	-0.778565	0.973155	-2.301509e-01	-0.212327	0.977199	-0.789905	-0.613230	0.184649	0.111501	3063.189961
242	2024-03-30	5228	-0.446983	0.894542	-0.600412	-0.799691	0.983732	-1.796449e-01	-0.279012	0.960288	-0.734304	-0.678820	0.196190	0.078053	2208.191865

	classic_bike_rentals	predictions_in_original_scale	residuals	ci_lower	ci_upper
0	13573	12359.100699	1213.899301	7647.044897	17071.156500
1	14901	11695.348159	3205.651841	6983.292358	16407.403961
2	14974	9904.524350	5069.475650	5192.468548	14616.580152
3	16130	11827.405121	4302.594879	7115.349319	16539.460922
4	8218	12999.873125	-4781.873125	8287.817323	17711.928927
...	...	...	...	...	...
238	3973	2844.847180	1128.152820	-1867.208621	7556.902982
239	4451	3437.817481	1013.182519	-1274.238321	8149.873283
240	5370	2906.787308	2463.212692	-1805.268493	7618.843110
241	4933	3063.189961	1869.810039	-1648.865840	7775.245763
242	5228	2208.191865	3019.808135	-2503.863937	6920.247666

	started_at	elect_bike_rentals	day	log_elect_bike_rentals	Differenced	diff_elect_bike_rentals
0	2020-07-13	8	Monday	2.079442	NaN	NaN
1	2020-07-14	14	Tuesday	2.639057	0.559616	6.0
2	2020-07-15	13	Wednesday	2.564949	-0.074108	-1.0
3	2020-07-28	31	Tuesday	3.433987	0.869038	18.0
4	2020-07-29	287	Wednesday	5.659482	2.225495	256.0
...	...	...	...	...	...	...
1340	2024-03-26	5208	Tuesday	8.557951	0.001153	6.0
1341	2024-03-27	5312	Wednesday	8.577724	0.019773	104.0
1342	2024-03-28	6564	Thursday	8.789355	0.211632	1252.0
1343	2024-03-29	5585	Friday	8.627840	-0.161516	-979.0
1344	2024-03-30	5273	Saturday	8.570355	-0.057485	-312.0

	started_at	prediction_5	prediction	prediction_95	predictions_in_original_scale
0	2023-07-03	0.063104	0.160729	0.264910	2981.881679
1	2023-07-04	-0.108034	0.048356	0.208856	897.813674
2	2023-07-05	-0.216313	-0.017359	0.181017	-320.943308
3	2023-07-06	-0.203877	0.026604	0.266109	494.406513
4	2023-07-07	-0.199672	0.068781	0.333747	1276.615190
...	...	...	...	...	...
267	2024-03-26	-1.908108	0.143813	2.253588	2668.161057
268	2024-03-27	-1.889099	0.168156	2.268470	3119.614864
269	2024-03-28	-1.845816	0.180898	2.275473	3355.940463
270	2024-03-29	-1.840256	0.184381	2.274342	3420.523764
271	2024-03-30	-1.952993	0.145655	2.240005	2702.312149

	started_at	elect_bike_rentals	fs_cos1	fs_sin1	fs_cos2	fs_sin2	fs_cos3	fs_sin3	fs_cos4	fs_sin4	fs_cos5	fs_sin5	scaled_elect_bike_rentals	prediction	predictions_in_original_scale
0	2023-07-03	13328	0.930818	-0.365484	0.732843	-0.680398	0.433469	-0.901169	0.074117	-0.997250	-0.295489	-0.955346	0.718592	0.160729	2981.881679
1	2023-07-04	13924	0.936967	-0.349418	0.755814	-0.654787	0.479378	-0.877609	0.142508	-0.989794	-0.212327	-0.977199	0.750728	0.048356	897.813674
2	2023-07-05	9192	0.942839	-0.333249	0.777890	-0.628401	0.524010	-0.851712	0.210225	-0.977653	-0.127594	-0.991826	0.495579	-0.017359	-320.943308
3	2023-07-06	12886	0.948432	-0.316982	0.799045	-0.601271	0.567248	-0.823547	0.276946	-0.960885	-0.041919	-0.999121	0.694759	0.026604	494.406513
4	2023-07-07	13875	0.953744	-0.300620	0.819255	-0.573430	0.608975	-0.793190	0.342357	-0.939570	0.044067	-0.999029	0.748086	0.068781	1276.615190
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
267	2024-03-26	5208	-0.473706	-0.880683	-0.551205	0.834370	0.995925	0.090190	-0.392347	-0.919817	-0.624210	0.781257	0.280761	0.143813	2668.161057
268	2024-03-27	5312	-0.458487	-0.888701	-0.579579	0.814916	0.989946	0.141444	-0.328176	-0.944617	-0.689017	0.724745	0.286369	0.168156	3119.614864
269	2024-03-28	6564	-0.443132	-0.896456	-0.607268	0.794497	0.981332	0.192322	-0.262452	-0.964945	-0.748730	0.662875	0.353877	0.180898	3355.940463
270	2024-03-29	5585	-0.427646	-0.903946	-0.634237	0.773138	0.970105	0.242687	-0.195486	-0.980707	-0.802907	0.596104	0.301089	0.184381	3420.523764
271	2024-03-30	5273	-0.412034	-0.911169	-0.660457	0.750864	0.956294	0.292406	-0.127594	-0.991826	-0.851148	0.524926	0.284266	0.145655	2702.312149

	elect_bike_rentals	predictions_in_original_scale	residuals	ci_lower	ci_upper
0	13328	2981.881679	10346.118321	-6201.070764	12164.834122
1	13924	897.813674	13026.186326	-8285.138769	10080.766116
2	9192	-320.943308	9512.943308	-9503.895751	8862.009135
3	12886	494.406513	12391.593487	-8688.545930	9677.358956
4	13875	1276.615190	12598.384810	-7906.337253	10459.567633
...	...	...	...	...	...
267	5208	2668.161057	2539.838943	-6514.791385	11851.113500
268	5312	3119.614864	2192.385136	-6063.337579	12302.567307
269	6564	3355.940463	3208.059537	-5827.011979	12538.892906
270	5585	3420.523764	2164.476236	-5762.428679	12603.476207
271	5273	2702.312149	2570.687851	-6480.640294	11885.264592

Forecasting Bike Share Usage in Chicago¶

Introduction¶

Project Content¶

Chicago Divvy Bicycle Dataset¶

1. Load Packages¶

2. Clean up Divvy Bicycle Data¶

Check 2020 Q1¶

Check 04/2020¶

Check 05/2020¶

Check 06/2020¶

Check 07/2020¶

Check 08/2020¶

Check 09/2020¶

Check 10/2020¶

Check 11/2020¶

Check 12/2020¶

Check 01/2021¶

Check 02/2021¶

Check 03/2021¶

Check 04/2021¶

Check 05/2021¶

Check 06/2021¶

Check 07/2021¶

Check 08/2021¶

Check 09/2021¶

Check 10/2021¶

Check 11/2021¶

Check 12/2021¶

Check 01/2022¶

Check 02/2022¶

Check 03/2022¶

Check 04/2022¶

Check 05/2022¶

Check 06/2022¶

Check 07/2022¶

Check 08/2022¶

Check 09/2022¶

Check 10/2022¶

Check 11/2022¶

Check 12/2022¶

Check 01/2023¶

Check 02/2023¶

Check 03/2023¶

Check 04/2023¶

Check 05/2023¶

Check 06/2023¶

Check 07/2023¶

Check 08/2023¶

Check 09/2023¶

Check 10/2023¶

Check 11/2023¶

Check 12/2023¶

Check 01/2024¶

Check 02/2024¶

Check 03/2024¶

Bike Data (01/01/2020 - 03/31/2024)¶

Keep where 'started_at' is between '2020-03-01' and '2024-03-31'¶

Categorical Rideable Type¶

Check Missing Value¶

Create a time series of bike trip counts¶

3. Load Divvy Bicyle Dataset¶

4. Exploratory Data Analysis¶

5. Data Transformation¶

5.1 Classic Bike Transformation¶

a. Classic Bike Log Transformation¶

b. Classic-Bike-Difference Transformation¶

5.2 Electric Bike Transformation¶

a. Electric Bike Log Transformation¶

b. Electric Bike Difference Transformation¶

6. Build Time Series Model & Evaluation¶

6.1 Calculate Cutoff Date¶

Classic Bike Cutoff Date¶

Electric Bike Cutoff Date¶

6.2 Training Testing Data¶

Classic Bike Split¶

Electric Bike Split¶

6.3 Long Short Term Memory¶

(a) LSTMs Classic Bike¶

Learning Curve (Classic Bike LSTM Model)¶

Evaluation Metrics (Classic Bike LSTM Model)¶